Interpretable Locomotion Prediction in Construction Using a Memory-Driven LLM Agent With Chain-of-Thought Reasoning

Chao Wang; Ehsan Ahmadi

REVIEW 4 major objections 5 minor 46 references

Interpretable Locomotion Prediction in Construction Using a Memory-Driven LLM Agent With Chain-of-Thought Reasoning

T0 review · 4 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Adding short-term and long-term memory to a language-model agent raises its ability to predict construction locomotion from spoken commands and smart-glasses video, lifting weighted F1 from 0.73 to 0.90 and sharply improving confidence…

desk verdict A useful application idea undermined by an ablation that lets the previous label leak through hard-coded transition rules; the 0.73-to-0.90 F1 story doesn't hold up as stated. read the letter →

arxiv 2504.15263 v1 pith:YJPGLGCG submitted 2025-04-21 cs.RO

Ehsan Ahmadi , Chao Wang This is my paper · ORCID

classification cs.RO

keywords locomotionpredictionexoskeletoncontrolintentrecognitionLLMagentsshort-termmemorylong-termchain-of-thoughtreasoningmultimodalperception

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that a locomotion prediction agent built around a large language model can predict a construction worker's movement mode (walking, climbing a ladder, stepping over an obstacle, etc.) from a spoken command and nine smart-glasses frames, and that the prediction becomes substantially better when the agent is given two kinds of memory: a short-term buffer of the most recent events and a long-term store of similar past situations. On a 226-sample dataset mixing clear, vague, and safety-critical commands, the authors report weighted F1 rising from 0.73 without memory to 0.81 with short-term memory and 0.90 with both memory systems. They also report that the model's self-assessed confidence becomes much better calibrated, with Brier score falling from 0.244 to 0.090 and expected calibration error from 0.222 to 0.044. The practical motivation is exoskeleton control: a wearable robot that knows whether the user is about to ascend a ladder or walk on level ground can assist at the right moment and avoid unsafe actions.

What carries the argument

The load-bearing mechanism is a clarity-gated refinement loop with discrepancy-weighted retrieval. The perception module first produces a prediction plus four self-reported scores: vagueness, discrepancy, importance, and confidence. A weighted clarity score decides whether the prediction is accepted or sent to a refinement module that re-prompts the model with memories retrieved from the long-term store. The retrieval itself is score-driven: text and image embeddings are blended with weights $1-d$ and $d$, where $d$ is the discrepancy between the spoken command and the visual frames, so that when words and vision conflict, the visual evidence dominates. Retrieved events are ranked by a composite score that prioritizes similarity, importance, and confidence while penalizing vagueness and discrepancy, with safety-critical events decaying more slowly and being penalized less. Short-term memory supplies the immediate safety context, including explicit rules about safe locomotion transitions, such as allowing a ladder descent only after a ladder ascent. Together these components turn the language model from an open-loop classifier into one that reconsiders its answer when its own scores say the input is ambiguous.

What would settle it

Collect a larger dataset of real construction workers performing the same twelve locomotion modes on actual sites, with independent ground-truth labels and human-annotated command-scene agreement, then compare the memory agent's F1 and calibration under the same configuration. If the discrepancy score stops separating correct from incorrect predictions, or if a no-memory baseline matches the memory agent when the command set is realistic rather than simulated, the central claim would be contradicted.

Watch

Extended reading notes

Core claim

The central discovery is that memory—not more powerful perception—accounts for most of the accuracy gain in this agent. Without memory, the perception module alone, using chain-of-thought prompting, reaches a weighted F1 of only 0.73 on the 226-sample test set. Injecting short-term memory of recent locomotion events (retained for 45 seconds) raises this to 0.81, and adding long-term memory of similar past experiences retrieved from a vector store raises it further to 0.90. The memory modules also drive the calibration improvements: the Brier score drops from 0.244 to 0.090 and ECE from 0.222 to 0.044. The authors further find that on the subset of ambiguous samples that trigger refinement, long-term memory cuts incorrect predictions by more than two-thirds, from 31 to 9 errors, and that safety-critical commands, where the user's words conflict with the visual scene, improve from F1 0.38 without memory to 0.72 with both memories.

Load-bearing premise

The central premise is that the 226 hand-collected samples, with simulated commands issued in a controlled environment, genuinely represent the variety and noise of real construction locomotion, and that the language model's self-reported vagueness, discrepancy, and confidence scores are accurate enough to drive retrieval and refinement decisions.

Editorial extensions

If this is right

On the paper's own ablation, every reported performance gain comes from memory: F1 rises 0.73→0.81 with short-term memory and 0.81→0.90 with long-term memory, so memory is the operative ingredient rather than the base LLM.
Calibration metrics move in the same direction, so the agent not only predicts more often correctly but its stated confidence tracks its accuracy more tightly—a property that matters for safety systems that act on probabilities.
The biggest improvements are on safety-critical commands, where the command contradicts the visual scene; the refinement module corrects most of these errors when long-term memory provides relevant similar events.
Some classes remain hard: vertical ladder climbing (F1 0.62–0.67) and obstacle stepping (0.86–0.92) still confuse with level-ground navigation, especially under vague or safety-critical commands.
The agent can override a misleading command (e.g., saying 'walk' while at the top of a ladder) by combining visual evidence with the recent context held in short-term memory.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The same design could be tested on other intent-prediction tasks where users give terse or misleading commands—assistive wheelchairs, teleoperation, or surgical robots—but the authors do not claim this extension.
A natural testable check is whether the improvement survives when retrieved memories are replaced with random past events matched for class and time: if random retrieval performs as well, the benefit would come from the refinement prompt rather than from genuinely similar history.
The discrepancy score is doing double duty (gating refinement, weighting text versus image embeddings, and penalizing LTM ranking); a hidden risk is that if the LLM's discrepancy estimate is biased, all three mechanisms skew together. That is not examined in the paper.
One could ablate the safety-transition rules in STM separately from the memory content; the rules encode domain knowledge, so removing them might separate the contribution of stored events from the contribution of the rule book.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Quick take: the idea—an LLM agent with short- and long-term memory for locomotion prediction from speech plus egocentric vision—is new for the exoskeleton/construction niche, and the system description is clean. But the headline ablation result is not trustworthy: the memory gain is largely explained by previous-label leakage and hard-coded transition rules in the prompt, not by the memory modules learning anything.

What the paper does well: the architecture is clearly laid out, related work on memory-augmented LLM agents is reasonably covered, and the decision to use the LLM's own vagueness/discrepancy/confidence scores to trigger refinement is a pragmatic idea. The class-level error analysis is honest about weak spots like vertical ladder and obstacle classes.

The soft spot is load-bearing. The perception prompt (Figure 2) includes deterministic rules: if the most recent memory says the user is climbing a construction ladder, the model must output ladder-down next; it also lists allowed safe transitions. Since the 226 samples are a continuous recording, STM gives the model the immediately preceding (usually correct) true label. NoMem doesn't have that, so part of the reported gain is just having the previous label plus a transition table. The paper omits the obvious control: a rule-based baseline that gets the same previous label and the same transition rules. It also lacks a held-out split for LTM (which is tested on the same samples it stores), error bars, and any external validation of the self-reported scores. The calibration improvement may be a side effect of the '100%' language in the prompt.

These issues don't make the concept worthless, but they mean the 0.73→0.90 F1 story is not established. The right next step is a revised evaluation with a rule-based baseline, a true train/test separation for LTM, and interval estimates. Who should read it: people working on exoskeleton intent recognition or LLM agents for physical tasks will find the application interesting, but should treat the numbers with caution. I'd still send this to a serious referee, because the area is relevant and the architecture is reusable, but with a strong request to fix the evaluation before acceptance.

Referee Report

4 major / 5 minor

Summary. The paper proposes an LLM-based locomotion prediction agent for exoskeleton assistance in construction. The system processes spoken commands and nine egocentric FOV frames through a gpt-4o Perception Module with chain-of-thought prompting, a Short-Term Memory (STM) for recent events, a Long-Term Memory (LTM) with vector retrieval, and a Refinement Module that re-prompts the LLM when a self-reported clarity score is low. On a 226-sample dataset with 12 locomotion modes and clear, vague, and safety-critical commands, the authors report weighted F1 of 0.73 (no memory), 0.81 (STM), and 0.90 (STM+LTM), with Brier Score falling from 0.244 to 0.090 and ECE from 0.222 to 0.044. The central claim is that adding memory systems improves prediction accuracy, calibration, and handling of safety-critical commands.

Significance. If the reported effects were robust, this would be a valuable demonstration that memory-augmented LLM agents can address locomotion intent recognition for exoskeletons in safety-critical, unstructured environments, going beyond the flat-ground walking tasks that dominate prior work. The paper deserves credit for publishing the full perception prompt (Figure 2), covering a multi-class dataset that includes ladders, obstacles, and low-space navigation, and reporting calibration metrics alongside F1. However, the central claim is not yet established: the ablation design does not separate the contribution of the memory modules from the contribution of the previous label and the hand-authored transition rules embedded in the prompt, and the evaluation lacks repeated runs and a held-out split. The paper's value is therefore conditional on a substantial revision of the evaluation protocol.

major comments (4)

[§4.4, Figure 2, §5.2] The ablation confounds memory with the availability of the previous label and with deterministic transition rules. In the memory conditions, STM stores the previous locomotion mode (§3.3.1), and the Perception prompt in Figure 2 instructs the model to apply rules such as 'If the most recent memory shows the user climbing a construction ladder, they will 100% climb down the same construction ladder' and to restrict transitions to a listed safe set. The NoMem condition has neither the previous label nor these rules. Since the 226 samples form a continuous sequence (§5.2), STM effectively provides the immediately preceding prediction (usually the correct ground-truth label) for most samples, so the reported F1 gain from 0.73 to 0.81 to 0.90 measures the joint effect of previous-label availability, hand-authored rules, and memory modules rather than the incremental value of memory. Section 6 discusses limitations for vertical ladders and obstacle classes but does not address this confound. The paper needs a deterministic rule-based baseline that receives the same previous label and the same transition table but no LLM or learned retrieval, and an ablation that removes the prompt rules while keeping memory, to establish the central claim.
[§4.2, §4.4, Tables 4–5] All reported metrics are point estimates from a single pass over 226 samples, with no error bars, no repeated runs, and no held-out split. This is particularly problematic because the Perception Module samples gpt-4o with temperature 0.7 (Table 3), making the results stochastic, and because the paper does not state whether any validation split was used before reporting the configuration in Table 3. The authors should report mean and standard deviation (or confidence intervals) over multiple independent runs, tune hyperparameters on a development set, and evaluate LTM generalization with memory built only from training samples and tested on unseen held-out samples.
[§3.4, §5.3] The refinement trigger and LTM ranking are driven entirely by the LLM's self-reported confidence, vagueness, and discrepancy scores, but these scores are never validated against external ground truth or human judgments. The analysis in §5.3 itself shows that confidence is almost non-discriminative between correct and incorrect predictions (means 0.940 vs 0.920), yet confidence still enters the clarity score with weight 0.2 and the LTM composite score with weight 0.15. The claimed calibration improvement (Brier Score 0.244→0.090, ECE 0.222→0.044) is therefore partly self-referential, because the model's own confidence is used both to decide which samples to refine and to compute the calibration metrics. The authors should calibrate the self-reported scores against an independent measure (e.g., human raters or a supervised classifier), or remove the self-referential score from the decision rule.
[§5.1, Figure 6] The attribution of error reduction to LTM in Figure 6 is not cleanly supported. The figure reports, for samples that undergo refinement in the full system, a shift from 17 correct/31 incorrect in NoMem to 39 correct/9 incorrect in STM+LTM. But the Refinement Module uses a different prompt at a different temperature with both STM and LTM context, so this comparison does not isolate LTM; moreover, the subset is selected by the full system's clarity threshold, which introduces selection on the outcome. The paper should report a matched comparison (e.g., STM+LTM with and without the Refinement Module) and full test-set confusion matrices for each condition before claiming that LTM 'reduces incorrect predictions by over two-thirds.'

minor comments (5)

[§4.1] The phrase 'within a environment tailored to simulate' should be 'within an environment'; the paper should also report the number of participants, the data collection protocol, and whether the spoken commands were scripted or spontaneous.
[Figure 2] The prompt display contains several garbled bullet characters and irregular sub-bullet formatting; please clean up the figure so that the exact prompt text is unambiguous.
[§1, §6] The paper repeatedly describes the system as 'interpretable', but no human evaluation of the chain-of-thought explanations is reported; the authors should state precisely what interpretability claim they are making and support it.
[§5.3] The claim that the score distributions 'justify' the clarity and LTM weights is post hoc, because those weights were already used to generate the scores; please rephrase or provide a validation-based justification.
[References] Some references are incomplete as printed; for example, [23] (MemGPT) lacks a publication venue and page numbers, and a few other entries have inconsistent formatting.

Circularity Check

1 steps flagged · score 6.0 of 10

Memory gain is largely forced by previous-label leakage and hard-coded transition rules in the Perception prompt, not by the memory modules.

other [Figure 2 (Perception Prompt, Safety Analysis) and Section 3.3.1 (STM); ablation results in Section 5.1]
"If the most recent memory shows the user climbing a construction ladder, they will 100% climb down the same construction ladder, and vice versa. ... The command is: "{command}" & Short-Term Memory context (starting from the most recent memories): "{stm_summary}" [Fig. 2] ... STM context is formatted as "At {timestamp}: {locomotion_mode} in a {environment} environment, interacting with a {primary_object}," derived from prior perception outputs [Sec. 3.3.1]."

The STM summary fed to the LLM is the immediately preceding locomotion mode from prior perception outputs. On the continuous 226-sample dataset, this is effectively the previous true label. The prompt then hard-codes a deterministic successor rule ('100% climb down the same construction ladder') and a list of safe transitions, so for ladder and other transition cases the current prediction is forced by the previous label plus a hand-authored rule. The NoMem ablation does not receive the previous label at all, so the reported gain (F1 0.73 to 0.90) measures the value of adding the previous label and the transition table, not the learned value of the memory modules.

full rationale

The paper's strongest evidence for its central claim is the ablation in Table 4, comparing NoMem, STMOnly, and STM+LTM. However, the Perception prompt in Figure 2 explicitly instructs the model to use the most recent memory as a deterministic transition trigger (e.g., ladder up implies 100% ladder down), and Section 3.3.1 confirms that STM context is the previous locomotion mode extracted from prior perception outputs. In a chronological dataset, this gives the memory conditions access to the previous true label, which the NoMem condition lacks. The ablation therefore compares a model with previous-label leakage plus hand-coded transition rules against a model without that input; the resulting F1 improvement is forced by construction rather than demonstrating emergent memory reasoning. The paper does not report a rule-based baseline that receives the same previous label and a transition table, and the LTM retrieval and refinement are driven by the LLM's own self-reported confidence, vagueness, and discrepancy scores, which are validated only against the system's own predictions. These are internal-validity concerns that undermine the causal attribution of the performance gain to the memory modules, warranting a partial circularity score. There is no self-citation load-bearing or uniqueness-theorem issue; the circularity is in the evaluation logic.

Assumptions & free parameters 7 free parameters · 4 assumptions · 0 invented entities

The system relies on many hand-set weights and thresholds; the evaluation is on a private dataset with no code or baselines, and the prompt encodes explicit memory-use rules that directly produce part of the reported improvement.

free parameters (7)

LTM similarity weight = 0.65
Hand-chosen weight in the composite retrieval score; no sensitivity analysis or tuning procedure is described.
LTM importance weight = 0.2
Hand-chosen; controls ranking influence of importance scores.
LTM confidence weight = 0.15
Hand-chosen; controls ranking influence of LLM confidence.
LTM discrepancy penalty = 0.3
Hand-chosen; penalizes high-discrepancy events during retrieval.
LTM vagueness penalty = 0.2
Hand-chosen; penalizes vague events during retrieval.
Clarity score weights and threshold = vagueness 0.3, discrepancy 0.5, confidence 0.2; threshold 0.35 to 0.75
Hand-set parameters that determine when refinement is triggered; directly affect the reported gains.
STM retention window = 45 seconds
Hand-chosen; affects which recent events are included as context.

assumptions (4)

domain assumption The 226-sample dataset is accurately labeled and representative of construction locomotion and command patterns.
All evaluations and conclusions rely on this private dataset; no collection protocol details or inter-rater validation are given.
domain assumption The LLM's self-reported confidence, vagueness, and discrepancy scores are reliable indicators of prediction correctness and are suitable for calibration.
Brier score and ECE are computed from these self-assessed scores; no external calibration check is provided.
ad hoc to paper The safe-transition rules encoded in the Perception prompt are correct and complete.
The prompt explicitly states 'If the most recent memory shows the user climbing a construction ladder, they will 100% climb down the same construction ladder...', so the memory benefit is partly prescribed by these rules.
domain assumption GPT-4o can reliably interpret egocentric FOV frames and commands for locomotion intent.
No error analysis of the perception module's visual or linguistic understanding is provided apart from final classifications.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Interpretable Locomotion Prediction in Construction Using a Memory-Driven LLM Agent With Chain-of-Thought Reasoning." pith.science (2026). https://pith.science/paper/YJPGLGCG

@misc{pith2026250415263,
  author       = {Pith},
  title        = {Pith review of: Interpretable Locomotion Prediction in Construction Using a Memory-Driven LLM Agent With Chain-of-Thought Reasoning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YJPGLGCG}},
  note         = {Machine review of arXiv:2504.15263}
}

read the original abstract

Construction tasks are inherently unpredictable, with dynamic environments and safety-critical demands posing significant risks to workers. Exoskeletons offer potential assistance but falter without accurate intent recognition across diverse locomotion modes. This paper presents a locomotion prediction agent leveraging Large Language Models (LLMs) augmented with memory systems, aimed at improving exoskeleton assistance in such settings. Using multimodal inputs - spoken commands and visual data from smart glasses - the agent integrates a Perception Module, Short-Term Memory (STM), Long-Term Memory (LTM), and Refinement Module to predict locomotion modes effectively. Evaluation reveals a baseline weighted F1-score of 0.73 without memory, rising to 0.81 with STM, and reaching 0.90 with both STM and LTM, excelling with vague and safety-critical commands. Calibration metrics, including a Brier Score drop from 0.244 to 0.090 and ECE from 0.222 to 0.044, affirm improved reliability. This framework supports safer, high-level human-exoskeleton collaboration, with promise for adaptive assistive systems in dynamic industries.

Figures

Figures reproduced from arXiv: 2504.15263 by the authors.

**Figure 1.** High-Level Overview of the Agent’s Workflow To enhance prediction accuracy, the Perception Module incorporates context from Short-Term Memory (STM). Recent events stored in STM, such as prior locomotion modes, environmental conditions, and objects or obstacles the user interacted with, inform the interpretation of the current command and frames for a more consistent and safer transition. For example, if the user wa… view at source ↗

**Figure 2.** Perception Prompt Used by the Perception Module 6 [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 3.** Examples of FOV frames [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗

Figures from the paper (6 more)

**Figure 4.** Figure 4: Distribution of Command Types in the Dataset applied to favor deterministic outputs, ensuring higher reliability when resolving ambiguities or safetycritical scenarios. Text embeddings are generated using the text-embedding-ada-002 model, while FOV frames are encoded …

**Figure 5.** Figure 5: Weighted F1-Score by Command Type NoMem Correct (17 samples) NoMem Incorrect (31 samples) STMOnly Correct (18 samples) STMOnly Incorrect (30 samples) STM+LTM Correct (39 samples) STM+LTM Incorrect (9 samples) [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]

**Figure 6.** Figure 6: Prediction Outcome Change from NoMem to STMOnly to STM+LTM (Only for Samples Undergoing Refinement Using LTM) 12 [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]

**Figure 7.** Figure 7: Overall Error Rate with Class-wise Breakdown Climbing (1 instance), Low Space Navigation (1 out of 3 correct) has 2 instances misclassified as Vertical Ladder Down Climbing, and Sitting Down (1 out of 2 correct) has 1 instance misclassified as Low Space Navigation. In …

**Figure 8.** Figure 8: Confusion Matrices of STM+LTM Predictions Across Clear, Vague, and Safety-Critical Commands (darker colors indicate higher prediction counts) 15 [PITH_FULL_IMAGE:figures/full_fig_p015_8.png]

**Figure 9.** Figure 9: Kernel Density Estimation of Perception Module Scores (confidence, discrepancy, and vagueness) Stratified by Prediction Correctness weight of 0.5 in the clarity score formula (𝑤𝑑 = 0.5, 𝑤𝑣 = 0.3, 𝑤𝑐 = 0.2), which determines whether events are transferred directly to LT…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

46 extracted references · 30 canonical work pages

[1]

Integrating exoskeletons in the construction sector: a systematic review of empirical evaluation tools and future directions

Mohamad Iyad Al-Khiami, Søren Munch Lindhard, and Søren Wandahl. Integrating exoskeletons in the construction sector: a systematic review of empirical evaluation tools and future directions. Engineering, Construction and Architectural Management, 2024

work page 2024
[2]

A multitask, multilingual, multimodal evaluation of chatgpt on reasoning, hallucination, and interactivity.arXiv preprint arXiv:2302.04023, 2023

YejinBang,SamuelCahyawĳaya,NayeonLee,WenliangDai,DanSu,BryanWilie,HolyLovenia, Ziwei Ji, Tiezheng Yu, Willy Chung, et al. A multitask, multilingual, multimodal evaluation of chatgpt on reasoning, hallucination, and interactivity.arXiv preprint arXiv:2302.04023, 2023

arXiv 2023
[3]

Reviewofcontrolstrategies for lower-limb exoskeletons to assist gait.Journal of NeuroEngineering and Rehabilitation, 18: 1–34, 2021

RomainBaud,AliRezaManzoori,AukeIjspeert,andMohamedBouri. Reviewofcontrolstrategies for lower-limb exoskeletons to assist gait.Journal of NeuroEngineering and Rehabilitation, 18: 1–34, 2021

work page 2021
[4]

Language models are few-shot learners.arXiv preprint arXiv:2005.14165, 2020

Tom B Brown. Language models are few-shot learners.arXiv preprint arXiv:2005.14165, 2020

arXiv 2005
[5]

Exoskeletonsforindustrialapplicationandtheirpotentialeffectsonphysicalworkload

Michiel P De Looze, Tim Bosch, Frank Krause, Konrad S Stadler, and Leonard W O’sullivan. Exoskeletonsforindustrialapplicationandtheirpotentialeffectsonphysicalworkload. Ergonomics, 59(5):671–681, 2016

work page 2016
[6]

Exoskeletons: Contribution to occupational health and safety.Bio- engineering, 10(9):1039, 2023

Omar Flor-Unda, Bregith Casa, Mauricio Fuentes, Santiago Solorzano, Fabián Narvaez-Espinoza, and Patricia Acosta-Vargas. Exoskeletons: Contribution to occupational health and safety.Bio- engineering, 10(9):1039, 2023

work page 2023
[7]

Speech-based human-exoskeleton interaction for lower limb motion planning

EddieGuo,ChristopherPerlette,MojtabaSharifi,LukasGrasse,MatthewTata,VivianKMushah- war, and Mahdi Tavakoli. Speech-based human-exoskeleton interaction for lower limb motion planning. In2024IEEE4thInternationalConferenceonHuman-MachineSystems(ICHMS) ,pages 1–6. IEEE, 2024

work page 2024
[8]

Ziheng Huang, Sebastian Gutierrez, Hemanth Kamana, and Stephen MacNeil. Memory sandbox: Transparentandinteractivememorymanagementforconversationalagents.In AdjunctProceedings of the36th AnnualACMSymposium onUser InterfaceSoftware andTechnology, pages1–3, 2023. 17

work page 2023

Show all 46 references

[9]

Kinematic-based locomotion mode recognition for power augmentation exoskeleton.International Journal of Advanced Robotic Systems, 14(5): 1729881417730321, 2017

Hongchul Kim, Young June Shin, and Jung Kim. Kinematic-based locomotion mode recognition for power augmentation exoskeleton.International Journal of Advanced Robotic Systems, 14(5): 1729881417730321, 2017

2017
[10]

Potential of exoskeleton technologies to enhance safety, health, and performance in construction: Industry perspectives and future re- search directions

Sunwook Kim, Albert Moore, Divya Srinivasan, Abiola Akanmu, Alan Barr, Carisa Harris- Adamson, David M Rempel, and Maury A Nussbaum. Potential of exoskeleton technologies to enhance safety, health, and performance in construction: Industry perspectives and future re- search di...

2019
[11]

Large language models are zero-shot reasoners.Advances in neural information processing systems, 35: 22199–22213, 2022

Takeshi Kojima, Shixiang Shane Gu, Machel Reid, Yutaka Matsuo, and Yusuke Iwasawa. Large language models are zero-shot reasoners.Advances in neural information processing systems, 35: 22199–22213, 2022

2022
[12]

Environment classificationforroboticlegprosthesesandexoskeletonsusingdeepconvolutionalneuralnetworks

Brokoslaw Laschowski, William McNally, Alexander Wong, and John McPhee. Environment classificationforroboticlegprosthesesandexoskeletonsusingdeepconvolutionalneuralnetworks. Frontiers in Neurorobotics, 15:730965, 2022

2022
[13]

Amultimodalframeworkbasedondeep belief network for human locomotion intent prediction.Biomedical Engineering Letters, pages 1–11, 2024

JiayiLi,JianhuaZhang,KexiangLi,JianCao,andHuiLi. Amultimodalframeworkbasedondeep belief network for human locomotion intent prediction.Biomedical Engineering Letters, pages 1–11, 2024

2024
[14]

Fusion of human gaze and machine vision for predicting intended locomotion mode.IEEE Transactions on Neural Systems and Reha- bilitation Engineering, 30:1103–1112, 2022

Minhan Li, Boxuan Zhong, Edgar Lobaton, and He Huang. Fusion of human gaze and machine vision for predicting intended locomotion mode.IEEE Transactions on Neural Systems and Reha- bilitation Engineering, 30:1103–1112, 2022

2022
[15]

Metaagents: Simulating interactions of human behav- iors for llm-based task-oriented coordination via collaborative generative agents.arXiv preprint arXiv:2310.06500, 2023

Yuan Li, Yixuan Zhang, and Lichao Sun. Metaagents: Simulating interactions of human behav- iors for llm-based task-oriented coordination via collaborative generative agents.arXiv preprint arXiv:2310.06500, 2023

2023 arXiv
[16]

Chain of thought empowers transformers to solve inherently serial problems.arXiv preprint arXiv:2402.12875, 2024

Zhiyuan Li, Hong Liu, Denny Zhou, and Tengyu Ma. Chain of thought empowers transformers to solve inherently serial problems.arXiv preprint arXiv:2402.12875, 2024

2024 arXiv
[17]

Agentsims: An open-sourcesandboxforlargelanguagemodelevaluation

Jiaju Lin, Haoran Zhao, Aochi Zhang, Yiting Wu, Huqiuyue Ping, and Qin Chen. Agentsims: An open-sourcesandboxforlargelanguagemodelevaluation. arXivpreprintarXiv:2308.04026 ,2023

2023 arXiv
[18]

Real-timelocomotionmoderecognitionandassistivetorquecontrol for unilateral knee exoskeleton on different terrains.IEEE/ASME Transactions on Mechatronics, 25(6):2722–2732, 2020

XiuhuaLiuandQiningWang. Real-timelocomotionmoderecognitionandassistivetorquecontrol for unilateral knee exoskeleton on different terrains.IEEE/ASME Transactions on Mechatronics, 25(6):2722–2732, 2020

2020
[19]

Pso-svm-based online locomotion mode identification for rehabilitation robotic exoskeletons.Sensors, 16(9):1408, 2016

Yi Long, Zhi-Jiang Du, Wei-Dong Wang, Guang-Yu Zhao, Guo-Qiang Xu, Long He, Xi-Wang Mao, and Wei Dong. Pso-svm-based online locomotion mode identification for rehabilitation robotic exoskeletons.Sensors, 16(9):1408, 2016

2016
[20]

Gaitactivityclassificationwithconvolutionalneuralnetworkusing lower limb angle measurement from inertial sensors.IEEE Sensors Journal, 2024

DavidMartínez-Pascual,JoséMCatalán,AndreaBlanco-Ivorra,MónicaSanchís,FranciscaArán- Ais,andNicolásGarcía-Aracil. Gaitactivityclassificationwithconvolutionalneuralnetworkusing lower limb angle measurement from inertial sensors.IEEE Sensors Journal, 2024

2024
[21]

Asurveyofsensorfusionmethodsinwearablerobotics

DomenNovakandRobertRiener. Asurveyofsensorfusionmethodsinwearablerobotics. Robotics and Autonomous Systems, 73:155–170, 2015

2015
[22]

Gpt-4v(ision) system card, 2023

OpenAI. Gpt-4v(ision) system card, 2023. URL https://cdn.openai.com/papers/GPTV_ System_Card.pdf. Accessed: 2025-03-19

2023
[23]

Memgpt: Towards llms as operating systems

CharlesPacker, VivianFang, Shishir_GPatil, KevinLin, SarahWooders, andJoseph_EGonzalez. Memgpt: Towards llms as operating systems. 2023. 18

2023
[24]

Generative agents: Interactive simulacra of human behavior

Joon Sung Park, Joseph O’Brien, Carrie Jun Cai, Meredith Ringel Morris, Percy Liang, and Michael S Bernstein. Generative agents: Interactive simulacra of human behavior. InProceedings of the 36th annual acm symposium on user interface software and technology, pages 1–22, 2023

2023
[25]

Ai deception: A survey of examples, risks, and potential solutions.Patterns, 5(5), 2024

Peter S Park, Simon Goldstein, Aidan O’Gara, Michael Chen, and Dan Hendrycks. Ai deception: A survey of examples, risks, and potential solutions.Patterns, 5(5), 2024

2024
[26]

Real-timehybridlocomotionmoderecogni- tionforlowerlimbwearablerobots

Andrea Parri, Kebin Yuan, Dario Marconi, Tingfang Yan, Simona Crea, Marko Munih, Raf- faeleMolinoLova,NicolaVitiello,andQiningWang. Real-timehybridlocomotionmoderecogni- tionforlowerlimbwearablerobots. IEEE/ASMETransactionsonMechatronics ,22(6):2480–2491, 2017

2017
[27]

Performance evaluation of lower limb exoskeletons: a systematic review.IEEE Transactions on Neural Systems and Rehabilitation Engineering, 28(7):1573–1583, 2020

David Pinto-Fernandez, Diego Torricelli, Maria del Carmen Sanchez-Villamanan, Felix Aller, KatjaMombaur,RobertoConti,NicolaVitiello,JuanCMoreno,andJoseLuisPons. Performance evaluation of lower limb exoskeletons: a systematic review.IEEE Transactions on Neural Systems and Rehab...

2020
[28]

Predictive locomotion mode recognition and accurate gait phase estimation for hip exoskeleton on various terrains.IEEE Robotics and Automation Letters, 7(3):6439–6446, 2022

Yuepeng Qian, Yining Wang, Chuheng Chen, Jingfeng Xiong, Yuquan Leng, Haoyong Yu, and Chenglong Fu. Predictive locomotion mode recognition and accurate gait phase estimation for hip exoskeleton on various terrains.IEEE Robotics and Automation Letters, 7(3):6439–6446, 2022

2022
[29]

Sayplan: Groundinglargelanguagemodelsusing3dscenegraphsforscalablerobottaskplanning

Krishan Rana, Jesse Haviland, Sourav Garg, Jad Abou-Chakra, Ian Reid, and Niko Suenderhauf. Sayplan: Groundinglargelanguagemodelsusing3dscenegraphsforscalablerobottaskplanning. arXiv preprint arXiv:2307.06135, 2023

2023 arXiv
[30]

Abhishek Sharma and Eric Rombokas. Improving imu-based prediction of lower limb kinematics in natural environments using egocentric optical flow.IEEE Transactions on Neural Systems and Rehabilitation Engineering, 30:699–708, 2022

2022
[31]

NoahShinn,FedericoCassano,AshwinGopinath,KarthikNarasimhan,andShunyuYao.Reflexion: Language agents with verbal reinforcement learning.Advances in Neural Information Processing Systems, 36:8634–8652, 2023

2023
[32]

Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context.arXiv preprint arXiv:2403.05530, 2024

Gemini Team, Petko Georgiev, Ving Ian Lei, Ryan Burnell, Libin Bai, Anmol Gulati, Garrett Tanzer, Damien Vincent, Zhufeng Pan, Shibo Wang, et al. Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context.arXiv preprint arXiv:2403.05530, 2024

2024 arXiv
[33]

In2023IEEEInternationalConferenceonRoboticsandAutomation(ICRA) ,pages10478–10482

OleksiiTsepa,RomanBurakov,BrokoslawLaschowski,andAlexMihailidis.Continuousprediction of leg kinematics during walking using inertial sensors, smart glasses, and embedded computing. In2023IEEEInternationalConferenceonRoboticsandAutomation(ICRA) ,pages10478–10482. IEEE, 2023

2023
[34]

Control strategies for active lowerextremityprostheticsandorthotics: areview

MichaelRTucker,JeremyOlivier,AnnaPagel,HannesBleuler,MohamedBouri,OlivierLambercy, José del R Millán, Robert Riener, Heike Vallery, and Roger Gassert. Control strategies for active lowerextremityprostheticsandorthotics: areview. Journalofneuroengineeringandrehabilitation , 12:...

2015
[35]

Language models don’t always say what they think: Unfaithful explanations in chain-of-thought prompting.Advances in Neural Information Processing Systems, 36:74952–74965, 2023

Miles Turpin, Julian Michael, Ethan Perez, and Samuel Bowman. Language models don’t always say what they think: Unfaithful explanations in chain-of-thought prompting.Advances in Neural Information Processing Systems, 36:74952–74965, 2023

2023
[36]

Integralreal-timelocomotionmoderecognitionbasedonga-cnnforlowerlimbexoskeleton

Jiaqi Wang, Dongmei Wu, Yongzhuo Gao, Xinrui Wang, Xiaoqi Li, Guoqiang Xu, and Wei Dong. Integralreal-timelocomotionmoderecognitionbasedonga-cnnforlowerlimbexoskeleton. Journal of Bionic Engineering, 19(5):1359–1373, 2022. 19

2022
[37]

A survey on large language model based autonomous agents

LeiWang,ChenMa,XueyangFeng,ZeyuZhang,HaoYang,JingsenZhang,ZhiyuanChen,Jiakai Tang, Xu Chen, Yankai Lin, et al. A survey on large language model based autonomous agents. Frontiers of Computer Science, 18(6):186345, 2024

2024
[38]

Recmind: Largelanguagemodelpoweredagent for recommendation.arXiv preprint arXiv:2308.14296, 2023

Yancheng Wang, Ziyan Jiang, Zheng Chen, Fan Yang, Yingxue Zhou, Eunah Cho, Xing Fan, XiaojiangHuang,YanbinLu,andYingzhenYang. Recmind: Largelanguagemodelpoweredagent for recommendation.arXiv preprint arXiv:2308.14296, 2023

2023 arXiv
[39]

Finetunedlanguagemodelsarezero-shotlearners

Jason Wei, Maarten Bosma, Vincent Y Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, AndrewMDai, andQuocVLe. Finetunedlanguagemodelsarezero-shotlearners. arXivpreprint arXiv:2109.01652, 2021

2021 arXiv
[40]

Chain-of-thought prompting elicits reasoning in large language models.Advances in neural information processing systems, 35:24824–24837, 2022

Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. Chain-of-thought prompting elicits reasoning in large language models.Advances in neural information processing systems, 35:24824–24837, 2022

2022
[41]

The rise and potential of large language model based agents: A survey.Science China Information Sciences, 68(2):121101, 2025

Zhiheng Xi, Wenxiang Chen, Xin Guo, Wei He, Yiwen Ding, Boyang Hong, Ming Zhang, Junzhe Wang, Senjie Jin, Enyu Zhou, et al. The rise and potential of large language model based agents: A survey.Science China Information Sciences, 68(2):121101, 2025

2025
[42]

AaronJYoungandLeviJHargrove.Aclassificationmethodforuser-independentintentrecognition for transfemoral amputees using powered lower limb prostheses.IEEE Transactions on Neural Systems and Rehabilitation Engineering, 24(2):217–225, 2015

2015
[43]

Artificial neural network-based activities classification, gait phase estimation, and prediction.Annals of biomedical engineering, 51(7):1471–1484, 2023

ShuangyueYu,JianfuYang,Tzu-HaoHuang,JunxiZhu,ChristopherJVisco,FarahHameed,Joel Stein, Xianlian Zhou, and Hao Su. Artificial neural network-based activities classification, gait phase estimation, and prediction.Annals of biomedical engineering, 51(7):1471–1484, 2023

2023
[44]

Explainability for large language models: A survey.ACM Transac- tions on Intelligent Systems and Technology, 15(2):1–38, 2024

Haiyan Zhao, Hanjie Chen, Fan Yang, Ninghao Liu, Huiqi Deng, Hengyi Cai, Shuaiqiang Wang, Dawei Yin, and Mengnan Du. Explainability for large language models: A survey.ACM Transac- tions on Intelligent Systems and Technology, 15(2):1–38, 2024

2024
[45]

Memorybank: Enhancing largelanguagemodelswithlong-termmemory

Wanjun Zhong, Lianghong Guo, Qiqi Gao, He Ye, and Yanlin Wang. Memorybank: Enhancing largelanguagemodelswithlong-termmemory. In ProceedingsoftheAAAIConferenceonArtificial Intelligence, volume 38, pages 19724–19731, 2024

2024
[46]

A novel motion intention recognition approach for soft exoskeleton via imu.Electronics, 9 (12):2176, 2020

Lu Zhu, Zhuo Wang, Zhigang Ning, Yu Zhang, Yida Liu, Wujing Cao, Xinyu Wu, and Chunjie Chen. A novel motion intention recognition approach for soft exoskeleton via imu.Electronics, 9 (12):2176, 2020. 20

2020

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [1]

Integrating exoskeletons in the construction sector: a systematic review of empirical evaluation tools and future directions

Mohamad Iyad Al-Khiami, Søren Munch Lindhard, and Søren Wandahl. Integrating exoskeletons in the construction sector: a systematic review of empirical evaluation tools and future directions. Engineering, Construction and Architectural Management, 2024

work page 2024

[2] [2]

A multitask, multilingual, multimodal evaluation of chatgpt on reasoning, hallucination, and interactivity.arXiv preprint arXiv:2302.04023, 2023

YejinBang,SamuelCahyawĳaya,NayeonLee,WenliangDai,DanSu,BryanWilie,HolyLovenia, Ziwei Ji, Tiezheng Yu, Willy Chung, et al. A multitask, multilingual, multimodal evaluation of chatgpt on reasoning, hallucination, and interactivity.arXiv preprint arXiv:2302.04023, 2023

arXiv 2023

[3] [3]

Reviewofcontrolstrategies for lower-limb exoskeletons to assist gait.Journal of NeuroEngineering and Rehabilitation, 18: 1–34, 2021

RomainBaud,AliRezaManzoori,AukeIjspeert,andMohamedBouri. Reviewofcontrolstrategies for lower-limb exoskeletons to assist gait.Journal of NeuroEngineering and Rehabilitation, 18: 1–34, 2021

work page 2021

[4] [4]

Language models are few-shot learners.arXiv preprint arXiv:2005.14165, 2020

Tom B Brown. Language models are few-shot learners.arXiv preprint arXiv:2005.14165, 2020

arXiv 2005

[5] [5]

Exoskeletonsforindustrialapplicationandtheirpotentialeffectsonphysicalworkload

Michiel P De Looze, Tim Bosch, Frank Krause, Konrad S Stadler, and Leonard W O’sullivan. Exoskeletonsforindustrialapplicationandtheirpotentialeffectsonphysicalworkload. Ergonomics, 59(5):671–681, 2016

work page 2016

[6] [6]

Exoskeletons: Contribution to occupational health and safety.Bio- engineering, 10(9):1039, 2023

Omar Flor-Unda, Bregith Casa, Mauricio Fuentes, Santiago Solorzano, Fabián Narvaez-Espinoza, and Patricia Acosta-Vargas. Exoskeletons: Contribution to occupational health and safety.Bio- engineering, 10(9):1039, 2023

work page 2023

[7] [7]

Speech-based human-exoskeleton interaction for lower limb motion planning

EddieGuo,ChristopherPerlette,MojtabaSharifi,LukasGrasse,MatthewTata,VivianKMushah- war, and Mahdi Tavakoli. Speech-based human-exoskeleton interaction for lower limb motion planning. In2024IEEE4thInternationalConferenceonHuman-MachineSystems(ICHMS) ,pages 1–6. IEEE, 2024

work page 2024

[8] [8]

Ziheng Huang, Sebastian Gutierrez, Hemanth Kamana, and Stephen MacNeil. Memory sandbox: Transparentandinteractivememorymanagementforconversationalagents.In AdjunctProceedings of the36th AnnualACMSymposium onUser InterfaceSoftware andTechnology, pages1–3, 2023. 17

work page 2023