Pith. sign in

REVIEW 3 major objections 4 minor 53 references

Action- and Language-Conditioned Video Assessment for Embodied Control

T0 review · 3 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A two-stage vision-language evaluator that summarizes action-conditioned video transitions and rates them on a 0–3 progress scale lifts average task completion from 29.3% to 57.8% in simulated household instruction-following, narrowing…

desk verdict Genuinely new two-stage VLM trajectory evaluator with a useful ablation story, but the Table 1 baseline comparison is not verifiable as written. read the letter →

arxiv 2608.08273 v1 pith:ORPTBQR5 submitted 2026-08-08 cs.RO cs.CV

Hwanhee Kim , Jaehyun Jang , Seungmin Cha , Hyeonseo Yun , Donghoon Lee , Chang D. Yoo This is my paper · ORCID
classification cs.ROcs.CV
keywords vision-languagemodelsvideoassessmenttrajectoryevaluationembodiedcontroltask-progressfeedbackAIALVAclosed-looppolicyoptimization
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

The paper claims that trajectory evaluation for embodied instruction-following should be treated as action- and language-conditioned video assessment: instead of matching a final frame to the instruction or comparing embedding similarities, an evaluator should read the whole trajectory in the light of the actions executed. ALVA does this by asking a frozen vision-language model to describe frame-to-frame visual changes conditioned on the executed action sequence, then asking it to rate that description against the instruction on a discrete 0–3 progress scale. In simulated ALFRED/AI2-THOR household tasks, using these scores as terminal feedback in a closed-loop IQL pipeline yields a 57.8% average maximum task-completion rate, versus 29.3% for the strongest embedding baseline (RoboCLIP) and 75.3% for a ground-truth oracle. The authors argue this supports action-conditioned video assessment as an interpretable, state-free feedback mechanism for embodied control.

What carries the argument

The machinery is a two-stage VLM querying pipeline. Stage 1, action-conditioned transition summarization, builds a composite visual prompt from sampled frames annotated with timestep numbers, pairs it with a text prompt listing the executed action names and the natural-language instruction, and asks a frozen vision-language model (Gemini 1.5 Pro in the main experiments) to describe the visual differences between consecutive timesteps in the context of those actions. Stage 2, task-progress assessment, feeds the resulting natural-language summary plus the instruction back to the VLM and asks for a single integer in {0,1,2,3}; the answer is accepted only if it is exactly one of those values. Long trajectories are split into 10-frame segments whose summaries are concatenated, and the discrete score is used directly as a terminal reward with intermediate rewards set to zero in an IQL off-policy optimization loop.

What would settle it

Re-run the closed-loop comparison with CLIP, R3M, and RoboCLIP receiving their native input representations instead of the shared ResNet-18 features, and with natural occlusion left in place; if their maximum task-completion rates approach or exceed ALVA's 57.8% average, the reported advantage is not attributable to action-conditioned video assessment. A second check is a direct one-stage VLM success-judgment baseline that reads the same composite frames without the action-conditioned summary.

Watch

Extended reading notes

Core claim

ALVA's central claim is that an action-conditioned transition summary is the load-bearing intermediate for judging task progress. The ablation shows that adding the executed action log raises offline evaluation F1 from 50.8 to 70.9, and adding timestep annotations brings it to 77.2, while the full evaluator keeps false-positive rates below 1% in all four tested workspaces. When the resulting 0–3 score is used as terminal reward in the same IQL policy-optimization loop, ALVA's maximum completion rate of 57.8% substantially outperforms the tested static image-text and embedding-similarity baselines and reduces the gap to the oracle. The paper's stated conclusion is therefore that action- and language-conditioned video assessment provides a useful, interpretable feedback signal without privileged simulator state, for the evaluated simulated household tasks and VLM backbones.

Load-bearing premise

The comparison rests on the assumption that all evaluators had a fair start: the baseline methods were fed through a shared image encoder rather than their own native input pipelines, and objects were kept clearly visible, so the measured advantage could partly be an interface or visibility effect rather than the value of action-conditioned assessment.

Editorial extensions

If this is right

  • If ALVA's feedback is responsible for the gain, closed-loop embodied instruction-following can be trained without hand-engineered reward functions or privileged simulator state.
  • ALVA's near-zero false-positive rate means failed trajectories rarely receive a success signal, reducing the risk that feedback reinforces incorrect behavior.
  • Action logs are the single largest contributor to assessment reliability, indicating that trajectory evaluators for embodied control should condition on the executed action sequence rather than the video alone.
  • The 10-frame segment length used in closed-loop experiments preserves most of the diagnostic benefit of the optimal 20-frame window while cutting inference cost.
  • Evaluative 0–3 scoring encodes degree of task progress, which is better suited to multi-step instruction following than pairwise preference feedback.

Reading between the lines

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

  • A testable extension not in the paper: thresholding 'success' at score 2 instead of 3 could recover some of the recall lost to ALVA's conservative under-scoring while keeping false positives low.
  • Because Stage 1 accounts for more than 90% of end-to-end latency in the paper's own profiling, a distilled or fine-tuned transition summarizer, or sparse frame selection, is the natural path toward real-time use.
  • Scoring each 10-frame segment independently would convert ALVA into a per-segment sub-task feedback signal, potentially giving the policy denser credit assignment than the trajectory-level terminal reward used here.
  • The paper's own comparison leaves open whether CLIP, R3M, and RoboCLIP would do better with their native input interfaces; testing that variant would either strengthen ALVA's claim or reveal an interface mismatch.
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.

Referee Report

3 major / 4 minor

Summary. The paper proposes ALVA, a two-stage trajectory evaluator for embodied instruction-following. ALVA uses a frozen vision-language model to first summarize frame-to-frame visual transitions conditioned on the executed action sequence and the language instruction, and then asks the model to assign a discrete progress score in {0,1,2,3} based on that summary. The authors evaluate ALVA in modified ALFRED/AI2-THOR household environments, reporting conservative false-positive behavior in offline trajectory assessment, a latency/accuracy trade-off across VLM backbones, and improved maximum task-completion rates when the ALVA score is used as terminal reward in an IQL policy-optimization pipeline compared with CLIP, R3M, and RoboCLIP baselines. The central claim is that action- and language-conditioned video assessment is a practical and interpretable reward source for the tested simulated tasks, with ALVA reaching 57.8% average maximum completion versus 29.3% for the best baseline and 75.3% for the oracle.

Significance. If the central comparison is valid, the paper makes a useful empirical contribution: it demonstrates that a two-stage VLM querying protocol with explicit action-conditioned transition summaries can serve as a trajectory-level reward signal that is more informative than final-frame matching or embedding-similarity baselines in simulated household tasks. The paper is commendably honest in its limitations section, explicitly acknowledging conservative scoring, prompt/backbone dependence, the controlled-visibility modification, and the lack of natural-occlusion and real-world validation. The ablation in Table 2 isolating temporal annotations and action logs is informative, and the per-stage latency profiling in Table A2 is practically useful. However, the significance is tempered by the unresolved ambiguity in the baseline input representation, the absence of stated seed counts for the closed-loop results, and the modified environment; these issues affect the load-bearing comparative claim and require clarification or additional experiments before the paper's conclusions can be accepted.

major comments (3)
  1. [Appendix A.2 and Section 5.2] The baseline comparison in Table 1 is compromised by an ambiguous input-representation statement. Appendix A.2 says: 'For downstream policy optimization and baseline methods, each observation is encoded using a frozen ResNet-18 pre-trained on ImageNet, resulting in a 512 × 7 × 7 visual feature representation.' This can be read as stating that the CLIP, R3M, and RoboCLIP evaluators receive ResNet-18 features rather than their native input representations, which would make the low scores in Table 1 a consequence of interface mismatch rather than a fair assessment of those methods. Section 5.2 describes CLIP as computing cosine similarity between the final visual observation and the instruction, R3M as measuring embedding similarity, and RoboCLIP as computing multimodal alignment between video and instruction; these methods are designed for their own encoder spaces. Please clarify whether the baseline evaluators used their native image/video encoders or a shared ResNet-18 encoder, and if the latter, rerun the closed-loop experiments with the baselines' native interfaces.
  2. [Section 5.5 and Table 1] Table 1 reports 'maximum task completion rates' achieved during training, but the manuscript does not state the number of independent runs or seeds used to compute the reported means and standard deviations. Reporting the maximum over an optimization run is sensitive to noise and to how frequently evaluation is performed, and without seed counts the reader cannot assess whether the large gaps between ALVA and the baselines are statistically meaningful. Please report the number of seeds, the standard protocol for evaluating during training, and preferably the final or mean-over-last-N evaluation performance in addition to the maximum.
  3. [Appendix A.2 and Section 7.1] The controlled-visibility modification is a load-bearing assumption for the central effectiveness claim. Appendix A.2 states that picked-up objects are kept clearly visible in the agent's view, and Section 7.1 acknowledges that this simplifies visual interpretation; because ALVA explicitly reasons over intermediate visual transitions, this modification may disproportionately favor ALVA over final-frame or embedding-similarity baselines. The paper does not provide evidence that the relative ranking of evaluators is preserved under natural occlusion. Please add an experiment with the original, unmodified ALFRED visibility conditions, or at minimum quantify how many of the 40 tasks involve object occlusion in the original environment and discuss how this could affect the comparison.
minor comments (4)
  1. [Section 6.2 and Table 2] The temporal-window analysis in Section 6.2 finds that a segment length of 20 gives the strongest diagnostic reliability, but the closed-loop experiments and Table 2 use a segment length of 10; please state explicitly which segment length is used for each reported result.
  2. [Table 3] The preference-based feedback comparison in Table 3 is not described in the methods; please specify how pairwise preferences are elicited from the VLM, how multiple comparisons are aggregated, and how the resulting feedback is used in the downstream IQL pipeline.
  3. [Figure 6] The x-axis labels in Figure 6 appear garbled ('0 50 75 0 1000 Total Time (s)'); please reformat the figure so the time axis is readable and consistent across panels.
  4. [Data Availability] The method depends on specific prompt templates and parsing rules, but only a single illustrative example is shown in Figure 2; please include the full prompts used for all ablated configurations in the appendix, since 'available from the corresponding author upon reasonable request' is not a reproducible artifact for a closed-loop result.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central comparison is an empirical benchmark evaluated against external ground-truth labels, not a derivation from fitted inputs.

full rationale

ALVA's central claim is an empirical comparison in simulated ALFRED/AI2-THOR tasks. The evaluator is a frozen, pre-trained VLM with no parameters fitted to the target result; the discrete score is produced by the two-stage querying in Eqs. (4)-(8) and is compared against environment-provided ground-truth completion labels (Section 5.3). In closed-loop optimization, the VLM score is used as terminal reward, but the reported task-completion rates (Table 1) are measured by the environment's ground-truth success signal, so the claim does not reduce to the evaluator's own output. Design choices such as k=4, segment length 10/20, and prompt phrasing are author-selected hyperparameters or explicit trade-off analyses (Section 6.2), not fitted values renamed as predictions. The Appendix A.2 statement that "each observation is encoded using a frozen ResNet-18 pre-trained on ImageNet" could raise a baseline-fairness concern for CLIP/R3M/RoboCLIP, but that is a potential confound in the empirical comparison, not a circular step: no equation or self-citation makes the baseline scores equivalent to ALVA's inputs. The paper contains no load-bearing self-citations. The derivation chain is therefore self-contained against external benchmarks, and no circular step was identified.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The method introduces no new physical or architectural entities; it uses an off-the-shelf VLM with prompt-level structure. The load-bearing assumptions are about the reliability of VLM summarization, the fairness of the environment modification, and the representativeness of the ALFRED-based evaluation.

free parameters (3)
  • segment_length = 10 frames (closed-loop); 20 frames (offline best)
    Chosen based on offline diagnostic analysis (Section 6.2) and inference cost; affects the accuracy-latency trade-off and therefore the closed-loop results.
  • progress_scale_size_k = 4
    Discrete rating scale 0..3 set by design in Section 3.2; defines the feedback signal and the binarization threshold.
  • binarization_threshold = r >= 3
    Used only for the confusion-matrix analysis in Section 5.3; directly affects the reported precision and recall.
assumptions (4)
  • domain assumption The VLM's stage-1 summaries accurately describe action-induced visual transitions from the composite image and action-name prompt.
    Central mechanism of ALVA; if the summaries are confabulated, the stage-2 rating is built on unreliable evidence (Sections 4.1 and 6.1).
  • ad hoc to paper The modified controlled-visibility environment preserves relative task difficulty across evaluators.
    Appendix A.2 ensures picked-up objects remain visible; this may favor a method that reasons over intermediate frames and is not quantified against natural occlusion.
  • domain assumption Ground-truth success labels from ALFRED are accurate and the 40 sampled two-subtask tasks are representative.
    All metrics and closed-loop success rates depend on this label; Section 5.1 and Appendix B.
  • domain assumption Baseline evaluators receive input representations that let them perform at their intended capacity.
    Section 5.2 describes native interfaces, but Appendix A.2 says observations for baselines are encoded with frozen ResNet-18; if so, baselines may be disadvantaged.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Action- and Language-Conditioned Video Assessment for Embodied Control." pith.science (2026). https://pith.science/paper/ORPTBQR5

@misc{pith2026260808273,
  author       = {Pith},
  title        = {Pith review of: Action- and Language-Conditioned Video Assessment for Embodied Control},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ORPTBQR5}},
  note         = {Machine review of arXiv:2608.08273}
}
read the original abstract

Vision-based embodied agents executing multi-step natural language instructions require feedback mechanisms that assess task progress over complete trajectories. Conventional approaches based on final-frame matching or continuous embedding similarity may overlook intermediate transitions that are necessary for determining whether an instruction has been completed. We propose ALVA (Action- and Language-Conditioned Video Assessment), a trajectory evaluator that conditions its assessment on visual observations, the executed action sequence, and the natural language instruction. The method uses a pre-trained vision-language model (VLM) in two stages: it first summarizes frame-to-frame visual transitions conditioned on the executed actions and then assesses the generated summary with respect to the instruction to produce a discrete trajectory-level progress score. In simulated 3D household environments, ALVA exhibits a conservative assessment pattern with near-zero false-positive rates. When used as terminal feedback for closed-loop policy optimization, it provides more effective feedback than the evaluated static image and embedding-based visual baselines and reduces the performance gap to a ground-truth oracle. These results support action- and language-conditioned video assessment as an interpretable feedback mechanism for the evaluated simulated embodied-control tasks.

Figures

Figures reproduced from arXiv: 2608.08273 by the authors.

Figure 1
Figure 1. Overview of ALVA. (a) A language-conditioned controller collects an action-conditioned video by executing control commands from visual observations under a natural language instruction. (b) A two-stage VLM trajectory evaluator first summarizes action-induced visual transitions from the collected stream and then estimates task progress with respect to the instruction. The snowflake symbol indicates that the pre-train… view at source ↗
Figure 2
Figure 2. Illustration of the two-stage VLM querying process used in ALVA. Blue text indicates task￾and trajectory-specific variables that are replaced according to the input. The snowflake symbol indicates that the pre-trained VLM is frozen and its parameters are not updated during control optimization. (a) In Stage 1, Action-Induced Transition Summarization, sampled visual observations are arranged into a composite visual p… view at source ↗
Figure 3
Figure 3. Representative egocentric views of the four unseen AI2-THOR floor plans used in the ALFRED-based experiments. 5.2. Baseline Evaluators To evaluate the proposed action- and language-conditioned assessment framework, we compare ALVA against three zero-shot visual evaluation baselines—CLIP, R3M, and Robo￾CLIP—and include a ground-truth oracle as an upper-bound reference. • CLIP [4]: A widely used vision-language founda… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Kitchen Bathroom Living Room Bedroom 250 200 150 100 50 0 Negative (0) Positive (1) Negative (0) Positive (1) Predicted Label True Label 250 200 150 100 50 0 Negative (0) Positive (1) Negative (0) Positive (1) Predicted Label True Label 250 200 150 100 50 0 Negative (0…
Figure 5
Figure 5. Figure 5: Impact of the temporal window size (segment length) on action-conditioned assessment reliability. The analysis demonstrates a system trade-off between contextual fragmentation (narrow windows) and visual input downsampling and increased inference cost (wide windows). T…
Figure 6
Figure 6. Figure 6: illustrates the trade-offs between total execution latency and evaluation accuracy. Lightweight VLMs have lower latency but exhibit lower accuracy and precision. Flagship-tier models provide higher reliability but require longer inference times. Crucially, our system l…
Figure 7
Figure 7. Figure 7: Closed-loop autonomous control performance across four simulated 3D embodied workspaces. The curves track the task completion rate over continuous control optimization it￾erations for Kitchen, Bathroom, Living Room, and Bedroom environments. The solid lines denote the …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 49 canonical work pages

  1. [1]

    Understanding natural language commands for robotic navigation and mobile manipulation

    Tellex, S.; Kollar, T.; Dickerson, S.; Walter, M.; Banerjee, A.; Teller, S.; Roy, N. Understanding natural language commands for robotic navigation and mobile manipulation. In Proceedings of the AAAI Conference on Artificial Intelligence, San Francisco, CA, USA, 7–11 August 2011

    work page 2011

  2. [2]

    Vision-and- language navigation: Interpreting visually-grounded navigation instructions in real environments

    Anderson, P .; Wu, Q.; Teney, D.; Bruce, J.; Johnson, M.; Sünderhauf, N.; Reid, I.; Gould, S.; Van Den Hengel, A. Vision-and- language navigation: Interpreting visually-grounded navigation instructions in real environments. InProceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR); IEEE: Piscataway, NJ, USA, 2018

    work page 2018

  3. [3]

    Alfred: A benchmark for interpreting grounded instructions for everyday tasks

    Shridhar, M.; Thomason, J.; Gordon, D.; Bisk, Y.; Han, W.; Mottaghi, R.; Zettlemoyer, L.; Fox, D. Alfred: A benchmark for interpreting grounded instructions for everyday tasks. InProceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR); IEEE: Piscataway, NJ, USA, 2020

    work page 2020

  4. [4]

    Learning transferable visual models from natural language supervision

    Radford, A.; Kim, J.W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P .; Clark, J.; et al. Learning transferable visual models from natural language supervision. In Proceedings of the International Conference on Machine Learning (ICML), Virtual, 18–24 July 2021

    work page 2021

  5. [5]

    Zero-shot reward specification via grounded natural language

    Mahmoudieh, P .; Pathak, D.; Darrell, T. Zero-shot reward specification via grounded natural language. In Proceedings of the International Conference on Machine Learning (ICML), Baltimore, MD, USA, 17–23 July 2022

    work page 2022

  6. [6]

    Vision-language models are zero-shot reward models for reinforcement learning

    Rocamonde, J.; Montesinos, V .; Nava, E.; Perez, E.; Lindner, D. Vision-language models are zero-shot reward models for reinforcement learning. In Proceedings of the International Conference on Learning Representations (ICLR), Vienna, Austria, 7–11 May 2024

    work page 2024

  7. [7]

    R3m: A universal visual representation for robot manipulation

    Nair, S.; Rajeswaran, A.; Kumar, V .; Finn, C.; Gupta, A. R3m: A universal visual representation for robot manipulation. In Proceedings of the Conference on Robot Learning (CoRL), Auckland, New Zealand, 14–18 December 2022

    work page 2022

  8. [8]

    Liv: Language-image representations and rewards for robotic control

    Ma, Y.J.; Kumar, V .; Zhang, A.; Bastani, O.; Jayaraman, D. Liv: Language-image representations and rewards for robotic control. In Proceedings of the International Conference on Machine Learning (ICML), Honolulu, HI, USA, 23–29 July 2023

    work page 2023

Show all 53 references
  1. [9]

    Roboclip: One demonstration is enough to learn robot policies

    Sontakke, S.; Zhang, J.; Arnold, S.; Pertsch, K.; Bıyık, E.; Sadigh, D.; Finn, C.; Itti, L. Roboclip: One demonstration is enough to learn robot policies. In Proceedings of the NeurIPS, Vancouver, BC, Canada, 10–15 December 2024. 19

    2024

  2. [10]

    GPT-4V(ision) System Card

    OpenAI. GPT-4V(ision) System Card. 2023. Available online: https://openai.com/index/gpt-4v-system-card/ (accessed on 3 August 2026)

    2023

  3. [11]

    Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context.arXiv2024, arXiv:2403.05530

    Reid, M.; Savinov, N.; Teplyashin, D.; Lepikhin, D.; Lillicrap, T.; Alayrac, J.b.; Soricut, R.; Lazaridou, A.; Firat, O.; Schrittwieser, J.; et al. Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context.arXiv2024, arXiv:2403.05530

    arXiv

  4. [12]

    Hello GPT-4o

    OpenAI. Hello GPT-4o. 2024. Available online: https://openai.com/index/hello-gpt-4o/ (accessed on 3 August 2026)

    2024

  5. [13]

    Ai2-thor: An interactive 3d environment for visual ai.arXiv2017, arXiv:1712.05474

    Kolve, E.; Mottaghi, R.; Han, W.; VanderBilt, E.; Weihs, L.; Herrasti, A.; Deitke, M.; Ehsani, K.; Gordon, D.; Zhu, Y.; et al. Ai2-thor: An interactive 3d environment for visual ai.arXiv2017, arXiv:1712.05474

    arXiv

  6. [14]

    Vision-language models for vision tasks: A survey.IEEE Trans

    Zhang, J.; Huang, J.; Jin, S.; Lu, S. Vision-language models for vision tasks: A survey.IEEE Trans. Pattern Anal. Mach. Intell.2024, 46, 5625–5644

    2024

  7. [15]

    Flamingo: A visual language model for few-shot learning

    Alayrac, J.B.; Donahue, J.; Luc, P .; Miech, A.; Barr, I.; Hasson, Y.; Lenc, K.; Mensch, A.; Millican, K.; Reynolds, M.; et al. Flamingo: A visual language model for few-shot learning. In Proceedings of the NeurIPS, New Orleans, LA, USA, 28 November–3 December 2022

    2022

  8. [16]

    Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution.arXiv2024, arXiv:2409.12191

    Wang, P .; Bai, S.; Tan, S.; Wang, S.; Fan, Z.; Bai, J.; Chen, K.; Liu, X.; Wang, J.; Ge, W.; et al. Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution.arXiv2024, arXiv:2409.12191

    arXiv

  9. [17]

    Rethinking spatiotemporal feature learning: Speed-accuracy trade-offs in video classification

    Xie, S.; Sun, C.; Huang, J.; Tu, Z.; Murphy, K. Rethinking spatiotemporal feature learning: Speed-accuracy trade-offs in video classification. InProceedings of the European Conference on Computer Vision (ECCV); Springer International Publishing: Cham, Switzerland, 2018

    2018

  10. [18]

    A survey of reinforcement learning informed by natural language

    Luketina, J.; Nardelli, N.; Farquhar, G.; Foerster, J.; Andreas, J.; Grefenstette, E.; Whiteson, S.; Rocktäschel, T. A survey of reinforcement learning informed by natural language. In Proceedings of the International Joint Conference on Artificial Intelligence (IJCAI), Macao,...

    2019

  11. [19]

    Reinforced cross-modal matching and self-supervised imitation learning for vision-language navigation

    Wang, X.; Huang, Q.; Celikyilmaz, A.; Gao, J.; Shen, D.; Wang, Y.F.; Wang, W.Y.; Zhang, L. Reinforced cross-modal matching and self-supervised imitation learning for vision-language navigation. InProceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recogn...

    2019

  12. [20]

    Calvin: A benchmark for language-conditioned policy learning for long-horizon robot manipulation tasks.IEEE Robot

    Mees, O.; Hermann, L.; Rosete-Beas, E.; Burgard, W. Calvin: A benchmark for language-conditioned policy learning for long-horizon robot manipulation tasks.IEEE Robot. Autom. Lett.2022,7, 7327–7334

    2022

  13. [21]

    Walk the talk: Connecting language, knowledge, and action in route instructions

    MacMahon, M.; Stankiewicz, B.; Kuipers, B. Walk the talk: Connecting language, knowledge, and action in route instructions. In Proceedings of the AAAI, Boston, MA, USA, 16–20 July 2006

    2006

  14. [22]

    Toward understanding natural language directions

    Kollar, T.; Tellex, S.; Roy, D.; Roy, N. Toward understanding natural language directions. InProceedings of the 2010 5th ACM/IEEE International Conference on Human-Robot Interaction (HRI); IEEE: Piscataway, NJ, USA, 2010

    2010

  15. [23]

    Tell Me Dave: Context-Sensitive Grounding of Natural Language to Manipulation Instructions

    Misra, D.K.; Sung, J.; Lee, K.; Saxena, A. Tell Me Dave: Context-Sensitive Grounding of Natural Language to Manipulation Instructions. In Proceedings of the Robotics: Science and Systems (RSS), Berkeley, CA, USA, 12–16 July 2014

    2014

  16. [24]

    Grounding English Commands to Reward Functions

    MacGlashan, J.; Babes-Vroman, M.; desJardins, M.; Littman, M.L.; Muresan, S.; Squire, S.; Tellex, S.; Arumugam, D.; Yang, L. Grounding English Commands to Reward Functions. In Proceedings of the Robotics: Science and Systems (RSS), Rome, Italy, 13–17 July 2015

    2015

  17. [25]

    Learning language-conditioned robot behavior from offline data and crowd-sourced annotation

    Nair, S.; Mitchell, E.; Chen, K.; Savarese, S.; Finn, C. Learning language-conditioned robot behavior from offline data and crowd-sourced annotation. In Proceedings of the Conference on Robot Learning (CoRL), Auckland, New Zealand, 14–18 December 2022

    2022

  18. [26]

    Algorithms for inverse reinforcement learning

    Ng, A.Y.; Russell, S. Algorithms for inverse reinforcement learning. In Proceedings of the International Conference on Machine Learning (ICML), Stanford, CA, USA, 29 June–2 July 2000

    2000

  19. [27]

    Apprenticeship learning via inverse reinforcement learning

    Abbeel, P .; Ng, A.Y. Apprenticeship learning via inverse reinforcement learning. In Proceedings of the Twenty-First ICML, Banff, AB, Canada, 4–8 July 2004; p. 1

    2004

  20. [28]

    Generative adversarial imitation learning

    Ho, J.; Ermon, S. Generative adversarial imitation learning. In Proceedings of the NeurIPS, Barcelona, Spain, 5–10 December 2016

    2016

  21. [29]

    From language to goals: Inverse reinforcement learning for vision-based instruction following

    Fu, J.; Korattikara, A.; Levine, S.; Guadarrama, S. From language to goals: Inverse reinforcement learning for vision-based instruction following. In Proceedings of the International Conference on Learning Representations (ICLR), New Orleans, LA, USA, 6–9 May 2019

    2019

  22. [30]

    Learning to Understand Goal Specifications by Modelling Reward

    Bahdanau, D.; Hill, F.; Leike, J.; Hughes, E.; Hosseini, S.; Kohli, P .; Grefenstette, E. Learning to Understand Goal Specifications by Modelling Reward. In Proceedings of the International Conference on Learning Representations (ICLR), Vancouver, BC, Canada, 30 April–3 May 2018

    2018

  23. [31]

    Reward Design with Language Models

    Kwon, M.; Xie, S.M.; Bullard, K.; Sadigh, D. Reward Design with Language Models. In Proceedings of the International Conference on Learning Representations (ICLR), Kigali, Rwanda, 1–5 May 2023

    2023

  24. [32]

    Language to rewards for robotic skill synthesis.arXiv2023, arXiv:2306.08647

    Yu, W.; Gileadi, N.; Fu, C.; Kirmani, S.; Lee, K.H.; Arenas, M.G.; Chiang, H.T.L.; Erez, T.; Hasenclever, L.; Humplik, J.; et al. Language to rewards for robotic skill synthesis.arXiv2023, arXiv:2306.08647. 20

    arXiv

  25. [33]

    Text2reward: Automated dense reward function generation for reinforcement learning

    Xie, T.; Zhao, S.; Wu, C.H.; Liu, Y.; Luo, Q.; Zhong, V .; Yang, Y.; Yu, T. Text2reward: Automated dense reward function generation for reinforcement learning. In Proceedings of the International Conference on Learning Representations (ICLR), Vienna, Austria, 7–11 May 2024

    2024

  26. [34]

    Eureka: Human-level reward design via coding large language models

    Ma, Y.J.; Liang, W.; Wang, G.; Huang, D.A.; Bastani, O.; Jayaraman, D.; Zhu, Y.; Fan, L.; Anandkumar, A. Eureka: Human-level reward design via coding large language models. In Proceedings of the International Conference on Learning Representations (ICLR), Vienna, Austria, 7–11...

    2024

  27. [35]

    Robogen: Towards unleashing infinite data for automated robot learning via generative simulation

    Wang, Y.; Xian, Z.; Chen, F.; Wang, T.H.; Wang, Y.; Fragkiadaki, K.; Erickson, Z.; Held, D.; Gan, C. Robogen: Towards unleashing infinite data for automated robot learning via generative simulation. In Proceedings of the International Conference on Machine Learning (ICML), Vie...

    2024

  28. [36]

    Cui, Y.; Niekum, S.; Gupta, A.; Kumar, V .; Rajeswaran, A. Can foundation models perform zero-shot task specification for robot manipulation? In Proceedings of the Learning for Dynamics and Control Conference, Stanford, CA, USA, 23–24 June 2022

    2022

  29. [37]

    Vision-language models as success detectors

    Du, Y.; Konyushkova, K.; Denil, M.; Raju, A.; Landon, J.; Hill, F.; de Freitas, N.; Cabi, S. Vision-language models as success detectors. In Proceedings of the Conference on Lifelong Learning Agents (CoLLAs), Montréal, QC, Canada, 22–25 August 2023

    2023

  30. [38]

    Rl-vlm-f: Reinforcement learning from vision language foundation model feedback

    Wang, Y.; Sun, Z.; Zhang, J.; Xian, Z.; Biyik, E.; Held, D.; Erickson, Z. Rl-vlm-f: Reinforcement learning from vision language foundation model feedback. In Proceedings of the International Conference on Machine Learning (ICML), Vienna, Austria, 21–27 July 2024

    2024

  31. [39]

    Deep reinforcement learning from human preferences

    Christiano, P .F.; Leike, J.; Brown, T.; Martic, M.; Legg, S.; Amodei, D. Deep reinforcement learning from human preferences. In Proceedings of the NeurIPS, Long Beach, CA, USA, 4–9 December 2017

    2017

  32. [40]

    Interactive learning from policy-dependent human feedback

    MacGlashan, J.; Ho, M.K.; Loftin, R.; Peng, B.; Wang, G.; Roberts, D.L.; Taylor, M.E.; Littman, M.L. Interactive learning from policy-dependent human feedback. In Proceedings of the International Conference on Machine Learning (ICML), Sydney, Australia, 6–11 August 2017

    2017

  33. [41]

    Learning reward functions from scale feedback

    Wilde, N.; Bıyık, E.; Sadigh, D.; Smith, S.L. Learning reward functions from scale feedback. In Proceedings of the Conference on Robot Learning (CoRL), London, UK, 8–11 November 2021

    2021

  34. [42]

    Rating-based reinforcement learning

    White, D.; Wu, M.; Novoseller, E.; Lawhern, V .J.; Waytowich, N.; Cao, Y. Rating-based reinforcement learning. In Proceedings of the AAAI, Vancouver, BC, Canada, 20–27 February 2024

    2024

  35. [43]

    NExT-QA: Next Phase of Question-Answering to Explaining Temporal Actions

    Xiao, J.; Shang, X.; Yao, A.; Chua, T.S. NExT-QA: Next Phase of Question-Answering to Explaining Temporal Actions. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition; IEEE: Piscataway, NJ, USA, 2021; pp. 9777–9786

    2021

  36. [44]

    From Representation to Reasoning: Towards Both Evidence and Commonsense Reasoning for Video Question-Answering

    Li, J.; Niu, L.; Zhang, L. From Representation to Reasoning: Towards Both Evidence and Commonsense Reasoning for Video Question-Answering. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition; IEEE: Piscataway, NJ, USA, 2022; pp. 21273–21282

    2022

  37. [45]

    Discovering the Real Association: Multimodal Causal Reasoning in Video Question Answering

    Zang, C.; Wang, H.; Pei, M.; Liang, W. Discovering the Real Association: Multimodal Causal Reasoning in Video Question Answering. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition; IEEE: Piscataway, NJ, USA, 2023; pp. 19027–19036

    2023

  38. [46]

    MECD: Unlocking Multi-Event Causal Discovery in Video Reasoning

    Chen, T.; Liu, H.; He, T.; Chen, Y.; Gan, C.; Ma, X.; Zhong, C.; Zhang, Y.; Wang, Y.; Lin, H.; et al. MECD: Unlocking Multi-Event Causal Discovery in Video Reasoning. In Proceedings of the Advances in Neural Information Processing Systems, Vancouver, BC, Canada, 10–15 December 2024

    2024

  39. [47]

    Open problems and fundamental limitations of reinforcement learning from human feedback.Trans

    Casper, S.; Davies, X.; Shi, C.; Gilbert, T.K.; Scheurer, J.; Rando, J.; Freedman, R.; Korbak, T.; Lindner, D.; Freire, P .; et al. Open problems and fundamental limitations of reinforcement learning from human feedback.Trans. Mach. Learn. Res.2023

    2023

  40. [48]

    Visual prompt tuning

    Jia, M.; Tang, L.; Chen, B.C.; Cardie, C.; Belongie, S.; Hariharan, B.; Lim, S.N. Visual prompt tuning. InProceedings of the European Conference on Computer Vision (ECCV); Springer Nature Switzerland: Cham, Switzerland, 2022

    2022

  41. [49]

    Visual prompting via image inpainting

    Bar, A.; Gandelsman, Y.; Darrell, T.; Globerson, A.; Efros, A. Visual prompting via image inpainting. In Proceedings of the NeurIPS, New Orleans, LA, USA, 28 November–3 December 2022

    2022

  42. [50]

    What does clip know about a red circle? visual prompt engineering for vlms

    Shtedritski, A.; Rupprecht, C.; Vedaldi, A. What does clip know about a red circle? visual prompt engineering for vlms. In Proceedings of the 2023 IEEE/CVF International Conference on Computer Vision(ICCV); IEEE: Piscataway, NJ, USA, 2023

    2023

  43. [51]

    Offline reinforcement learning with implicit q-learning

    Kostrikov, I.; Nair, A.; Levine, S. Offline reinforcement learning with implicit q-learning. In Proceedings of the International Conference on Learning Representations (ICLR), Virtual, 25–29 April 2022

    2022

  44. [52]

    Bootstrap your own skills: Learning to solve new tasks with large language model guidance

    Zhang, J.; Zhang, J.; Pertsch, K.; Liu, Z.; Ren, X.; Chang, M.; Sun, S.H.; Lim, J.J. Bootstrap your own skills: Learning to solve new tasks with large language model guidance. In Proceedings of the Conference on Robot Learning (CoRL), Atlanta, GA, USA, 6–9 November 2023

    2023

  45. [53]

    Deep residual learning for image recognition

    He, K.; Zhang, X.; Ren, S.; Sun, J. Deep residual learning for image recognition. InProceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition(CVPR); IEEE: Piscataway, NJ, USA, 2016. 21

    2016

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.