Pith. sign in

REVIEW 3 major objections 5 minor 55 references

Deep Reinforcement Learning for Dynamic Battery Management of Autonomous Order Pickers

T0 review · 3 major / 5 minor · reviewed 2026-07-11 · grok-4.5

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

Pith's one-line read A multi-agent PPO policy that jointly decides when, where, and how long warehouse robots charge raises order-completion rates by up to 6% over the best fixed-rule and DRL baselines while cutting recharging time.

desk verdict Solid applied multi-agent RL for joint charging/depot decisions in multi-block warehouses; the ~6% completion lift is directionally supported but rests on single-run point estimates without seed variance. read the letter →

arxiv 2607.05683 v1 pith:SOM3ZNNH submitted 2026-07-06 cs.LG cs.MAmath.OC

Taniya Shaji , Abhay Sobhanan , Christof Defryn This is my paper · ORCID
classification cs.LGcs.MAmath.OC
keywords warehouselogisticsdeepreinforcementlearningbatterychargingautonomousmobilerobotsproximalpolicyoptimizationmulti-agentsystemsaverage-rewardRLorderpicking
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

Warehouse robots keep stopping work to recharge, and simple battery thresholds plus nearest-station rules leave stations congested and orders unfinished when arrivals are random. This paper shows that a shared-actor, independent-critic PPO controller can learn, from local observations of battery, queues, neighbors, and distances, four linked decisions at once: whether to charge now, which fixed station to join, when to stop charging short of full, and when to return to the depot. The controller is trained with average reward rather than discounted return so every future order counts equally. In four- and six-block simulated warehouses the learned policy finishes up to 6% more orders than the strongest fixed-threshold, high-low, DQN, CTDE-PPO, and prior masked-PPO baselines, while spending less total time charging and waiting. The same policy remains superior under piecewise-constant real arrival patterns and under longer test horizons, and SHAP plus mean-state analysis recover transparent operating rules that managers can adopt even without running the network.

What carries the argument

Average-reward Independent PPO (shared actor + per-agent local critics) with differential GAE, action masks, and a sparse reward of +20 per pick start minus 1 per time-step; the agent observes its own battery/capacity/distances plus neighbors’ batteries and station queues so the policy can coordinate without a joint critic.

What would settle it

Re-run the identical multi-block layouts with payload-dependent drain and a simple collision or path-conflict model; if the 6% completion gap over Fixed-100-15 / Fixed-100-20 disappears or reverses, the central claim fails under realistic kinematics.

Watch

Extended reading notes

Core claim

When multi-block AMRs with shared fixed chargers must serve stochastic Poisson orders, an average-reward Independent PPO policy that jointly chooses station, charging duration (including early stop), and depot return raises order-completion rates by up to 6% over the strongest fixed-threshold and DRL baselines while simultaneously lowering charging and queue time per completed order.

Load-bearing premise

Robots never collide, battery drains only while moving and at a constant rate independent of load, and the minimum battery threshold is always enough to reach the nearest charger from anywhere in the block.

Editorial extensions

If this is right

  • Fixed lower/upper battery thresholds and nearest-station rules are systematically suboptimal once queues and stochastic arrivals matter.
  • Allowing early termination of charging when a neighbor’s battery is low measurably reduces station congestion and raises fleet throughput.
  • Depot trips should be deferred until capacity is nearly exhausted; the learned policy recovers this batching rule automatically.
  • Average-reward training (no discount factor) is necessary for long-horizon order-completion objectives; discounted PPO plateaus lower.
  • The extracted decision rules (charge near 25%, prefer nearest free station, stop early for low-battery neighbors) can be hard-coded as improved heuristics even without online DRL.

Reading between the lines

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

  • The same observation-and-mask design should transfer to other shared-resource fleets (AGVs, drones, last-mile robots) whose downtime is dominated by queuing rather than pure travel.
  • Because the policy already conditions on neighbor battery, adding a light communication channel for planned charge times could further cut residual queue waste without enlarging the action space.
  • If load-dependent drain is introduced later, the critic’s value estimates will shift mainly through the battery feature; retraining only the later layers may be sufficient.
  • Positioning chargers nearer the farthest blocks is a low-cost layout change that the SHAP analysis already flags as high-leverage.
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 / 5 minor

Summary. The paper formulates dynamic battery management for multi-block warehouse AMRs under Poisson order arrivals as a multi-agent MDP and solves it with average-reward Independent PPO (shared actor, per-agent critics, differential GAE, action masks). Agents jointly learn when to charge, which fixed station to use, when to stop charging, and when to return to the depot. Experiments on 4- and 6-block layouts compare IPPO to an adapted Bischoff masked-PPO baseline, DQN, CTDE PPO, FixedThreshold, and HighLow heuristics. The central empirical claim is up to ~6 percentage-point higher order-completion rates versus the strongest baseline, with lower charge and queue time; SHAP and mean-state analyses are used to interpret the learned policy, and sensitivity checks cover train/test horizons and time-varying arrivals.

Significance. Battery coordination under shared fixed chargers is a practically important and still under-modeled piece of warehouse AMR control. Strengths include a carefully specified MDP with feasibility masks (Section 4), a well-motivated average-reward objective with differential GAE (Section 5.1–5.3; Figure 2), a transparent IPPO training loop (Algorithm 1), public code, multi-baseline comparisons including a literature DRL adaptation, and useful SHAP/operational-rule interpretation (Section 6.3). If the reported gains hold under proper statistical validation and the stated modeling assumptions, the work is a solid applied contribution to DRL for warehouse logistics and offers transferable decision heuristics for practitioners.

major comments (3)
  1. Abstract and Tables 8–9 / Figure 7 report order-completion gains of up to ~6 points (e.g., 76 vs 70 in E-I µ=0.6; 60 vs 55 in E-II µ=0.9) as single point estimates from one training trajectory and one evaluation horizon per cell. No multi-seed means, standard deviations, or significance tests are provided. Multi-agent PPO is known to be seed- and entropy-schedule sensitive; without that evidence the headline superiority claim is not statistically secured. Please re-run evaluation (and ideally training) over multiple seeds and report mean±std (or confidence intervals) for completion rate, charge time, and wait time.
  2. Section 3.3 and Eqs. (1)–(3) assume collision-free motion and payload-independent constant depletion η, with b_min always sufficient to reach a station. These assumptions are shared by all in-simulator baselines, so they do not invalidate relative rankings inside the study, but they are load-bearing for any claim of operational transfer. The paper should either (i) stress-test relative ranking under mild collision costs or load-dependent drain, or (ii) clearly bound the claim to the idealized simulator and strengthen the limitations/future-work discussion so the Abstract does not over-generalize.
  3. Hyperparameters (Table 7b) are tuned via Optuna on E-I only (Figure 3) and then reused for E-II and dynamic-arrival tests. Given that the strongest baseline sometimes changes with layout and µ, please either retune (or at least ablate) key PPO/GAE settings on E-II, or show that the ranking is stable under a small hyperparameter perturbation so the cross-environment robustness claim is not confounded by E-I-specific tuning.
minor comments (5)
  1. Table 7(a) lists “γ” with values 0.50/0.60 and 0.75/0.90; the text and figures use these as order-arrival rates µ, and the method is average-reward (no discount). Fix the symbol so γ is not confused with the discarded discount factor.
  2. Abstract says the model learns “two key decisions” (station and duration), while the contributions and action space also include when to charge and when to go to the depot. Align the Abstract wording with Section 1 contributions.
  3. Figures 5 and 7–9 would benefit from error bars (once multi-seed runs exist) and consistent axis labels; several panels show blank “=” where µ should appear (extraction/typesetting issue).
  4. Clarify in Section 6.2 exactly which Bischoff reward/interrupt/A_full settings were selected as “best” and whether any re-tuning of that baseline’s thresholds was allowed, so the comparison is fully reproducible.
  5. Minor prose: “deport” appears once for “depot” in Section 6.3; check consistency of CS vs charging station notation.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: empirical multi-agent RL gains are measured against external baselines on held-out simulation, not forced by definition or self-citation.

full rationale

The paper formulates a multi-agent MDP (state, masked action space of size M+6, transitions for position/battery/capacity/queues in Eqs. 1–3, and a simple shaped reward of +20 pick / −1 step) and trains average-reward IPPO (shared actor, independent critics, differential GAE, running baseline ρ̄). Performance claims (order-completion rates up to 6 pp higher than FixedThreshold / Bischoff / DQN / CTDE-PPO, lower charge and queue time) are obtained by rolling out the learned policy on 8-hour test horizons and comparing point estimates in Tables 8–9 and Figures 7–9. The reward does not hard-code the completion-rate metric; the average-reward baseline is a standard stabilization device, not an identity that makes the reported gain tautological. Hyperparameters are tuned by Optuna on training episodes; evaluation uses separate longer horizons and additional sensitivity settings (dynamic µ, depot relocation, train/test duration). No uniqueness theorem, fitted constant renamed as prediction, or load-bearing self-citation appears. The work is therefore self-contained empirical RL; residual statistical concerns (single-seed tables) are correctness/robustness issues, not circularity.

Assumptions & free parameters 5 free parameters · 7 assumptions · 0 invented entities

The central performance claim rests on a simulated warehouse MDP plus many tuned RL and environment constants, not on free-standing theory. Domain assumptions (Poisson orders, collision-free motion, constant drain, always-reachable stations, local observability of neighbors) define the world in which the 6 pp gain is measured. Free parameters include Optuna-tuned PPO/GAE rates and hand-set environment rates (η, β, b_min, rewards). No new physical entities are postulated; “IPPO with average reward” and SHAP are tools, not invented ontology.

free parameters (5)
  • lr_actor / lr_critic
    Learning rates chosen by Bayesian Optuna search on E-I; final values 1.20e-4 and 4.86e-4 (Table 7). Policy quality depends on these fits.
  • PPO clip ε, GAE λ, entropy H0, baseline α_r̄
    Tuned in ranges (ε=0.208, λ=0.98, H0=0.164, α_r̄=0.0117). Control exploration/stability of the reported policies.
  • R_pick=+20, R_step=-1
    Hand-designed reward shaping; authors note more complex rewards underperformed. Completion-rate gains are relative to this choice.
  • η, β, b_min, b_max, K, V
    Environment battery and capacity constants (e.g., η=1, β=2, b_min=15 or 20, K=10). Set by authors, not estimated from real robot telemetry in the paper.
  • Entropy decay horizon 7000 episodes
    Empirically chosen schedule for α_H (Eq. 13); affects exploration vs exploitation balance during the 10k-episode training used for claims.
assumptions (7)
  • domain assumption Order arrivals follow a Poisson process (or piecewise-constant Poisson for sensitivity) with orders assigned uniformly to blocks.
    Section 3 and Table 3; defines the stochastic demand the policy is trained and tested against.
  • domain assumption Battery depletes only during transit at constant rate η; idle/queue/charge periods do not drain; recharge is constant β when first in queue.
    Section 3.3 and Eqs. (2)–(3). Load-independent drain is a strong simplification flagged by authors for future work.
  • domain assumption Agents are collision-free; Euclidean travel with linear interpolation is an adequate motion model.
    Section 3.3 Assumptions and Eq. (1). Removes path conflicts that dominate real multi-AMR systems.
  • domain assumption b_min is always sufficient to reach the nearest charging station from any location in the agent’s range.
    Section 3.3; prevents dead robots and simplifies feasibility of go-to-CS actions.
  • domain assumption Each block is exclusively serviced by one agent; stations are shared; agents observe neighbors’ battery, capacity, service mode, and relative distances locally.
    Sections 3–4; multi-agent coordination channel without centralized critic (IPPO).
  • ad hoc to paper Average-reward (undiscounted) objective is the appropriate long-run criterion for continuing warehouse throughput.
    Section 5.1 motivates via Mu et al. and Figure 2 comparison; standard discounted PPO is set aside by design choice.
  • domain assumption Independent PPO with shared actor and per-agent critics is a valid multi-agent learning scheme for this non-stationary environment.
    Section 5 methodology; known MARL approximation, not proved optimal for the warehouse game.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep Reinforcement Learning for Dynamic Battery Management of Autonomous Order Pickers." pith.science (2026). https://pith.science/paper/SOM3ZNNH

@misc{pith2026260705683,
  author       = {Pith},
  title        = {Pith review of: Deep Reinforcement Learning for Dynamic Battery Management of Autonomous Order Pickers},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SOM3ZNNH}},
  note         = {Machine review of arXiv:2607.05683}
}
read the original abstract

Battery charging of Autonomous Mobile Robots (AMRs) in warehouses is a critical operational challenge that heavily impacts both order processing times and throughput. In this study, we address the dynamic AMR charging problem under stochastic order arrivals, where robots must learn optimal charging decisions. Traditional fixed-rule heuristics often prove suboptimal in dynamic environments and fail to account for multi-AMR coordination, leading to severe resource inefficiencies. To overcome these limitations, we propose a Proximal Policy Optimization (PPO)-based Deep Reinforcement Learning (DRL) framework designed for multi-block warehouses with fixed charging stations. Our model dynamically learns two key decisions: charging station selection and optimal charging duration, explicitly accounting for anticipated queuing times at the stations. Extensive numerical experiments benchmark the proposed model against state-of-the-art DRL and traditional heuristic approaches. Results demonstrate that our PPO framework increases order-completion rates by up to 6\% compared to the strongest baseline, while significantly reducing the total time dedicated to recharging operations. Furthermore, we validate the model's robustness across diverse warehouse configurations and stochastic arrival rates. Finally, we interpret the learned DRL policy, offering valuable operational insights into its superiority over standard benchmarks.

Figures

Figures reproduced from arXiv: 2607.05683 by the authors.

Figure 1
Figure 1. Layout for a warehouse with 4 blocks, 3 aisles per block, 3 storage locations per rack and [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Order completion rate when using average reward versus discounted reward in a repre [PITH_FULL_IMAGE:figures/full_fig_p012_2.png] view at source ↗
Figure 3
Figure 3. Hyperparameter tuning results performed in E-I environment [PITH_FULL_IMAGE:figures/full_fig_p018_3.png] view at source ↗
Figures from the paper (12 more)
Figure 4
Figure 4. Figure 4: Propagation of actor and critic losses during training under different scenarios. [PITH_FULL_IMAGE:figures/full_fig_p019_4.png]
Figure 5
Figure 5. Figure 5: Reward, entropy, and order completion rates under different scenarios. [PITH_FULL_IMAGE:figures/full_fig_p020_5.png]
Figure 6
Figure 6. Figure 6: Order completion rate across agents in E-I environment with [PITH_FULL_IMAGE:figures/full_fig_p021_6.png]
Figure 7
Figure 7. Figure 7: Performance across different benchmarking algorithms [PITH_FULL_IMAGE:figures/full_fig_p022_7.png]
Figure 8
Figure 8. Figure 8: Percentage of time spent charging and waiting in queue at CS in different scenarios [PITH_FULL_IMAGE:figures/full_fig_p023_8.png]
Figure 9
Figure 9. Figure 9: Time spent charging for per completed order in different scenarios [PITH_FULL_IMAGE:figures/full_fig_p023_9.png]
Figure 10
Figure 10. Figure 10: Policy learned in E-I environment with µ = 0.60 6.3.1 SHAP Value Analysis A policy analysis is conducted using SHapley Additive exPlanations (SHAP) (Lundberg and Lee, 2017). SHAP identifies the state features in the observation space that most strongly influence the a…
Figure 11
Figure 11. Figure 11: Policy learnt in E-II environment with µ = 0.90 only under specific environmental conditions. For example, stop charging is available only when the agent is at a CS and first in the queue. To characterize the average policy behavior, 100 state observations are sampled…
Figure 12
Figure 12. Figure 12: SHAP values for different actions based on policy learned for E-I environment with [PITH_FULL_IMAGE:figures/full_fig_p026_12.png]
Figure 13
Figure 13. Figure 13: Order completion rate at different training and testing time in E-II environment [PITH_FULL_IMAGE:figures/full_fig_p029_13.png]
Figure 14
Figure 14. Figure 14: Dynamic arrival rate Order Completion Rate 0 10 20 30 40 50 60 70 80 90 100 IPPO Dynamic γ Fixed 100_20 Dynamic γ IPPO Constant γ Fixed 100_20 Constant γ overall A1 A2 A3 A4 A5 A6 [PITH_FULL_IMAGE:figures/full_fig_p031_14.png]
Figure 15
Figure 15. Figure 15: Effect of dynamic arrival rate based deep reinforcement learning framework is proposed to learn charging strategies instead of relying on fixed rules, which are currently dominant in the industry. The key decisions learned include deciding when to start and stop charg…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 55 canonical work pages

  1. [1]

    2025 , howpublished =

    Dresser, Scott , title =. 2025 , howpublished =

    work page 2025

  2. [2]

    Amazon has more than 1 million robots that sort, lift, and carry packages—see them in action , howpublished =

    Greenawalt, Tom , year =. Amazon has more than 1 million robots that sort, lift, and carry packages—see them in action , howpublished =

    work page

  3. [3]

    Nature , volume=

    Mnih, Volodymyr and others , title=. Nature , volume=

    work page

  4. [4]

    International Journal of Production Research , volume =

    Marcel Panzer and Benedict Bender , title =. International Journal of Production Research , volume =. 2022 , publisher =

    work page 2022

  5. [5]

    Coordinating recharging of large scale robotic teams , year=

    Drenner, Andrew and Janssen, Michael and Papanikolopoulos, Nikolaos , booktitle=. Coordinating recharging of large scale robotic teams , year=

    work page

  6. [6]

    2021 , issn =

    Research on online scheduling and charging strategy of robots based on shortest path algorithm , journal =. 2021 , issn =

    work page 2021

  7. [7]

    Procedia Manufacturing , volume =

    Raffaele Carli and Salvatore Digiesi and Mariagrazia Dotoli and Francesco Facchini , title =. Procedia Manufacturing , volume =. 2020 , note =

    work page 2020

  8. [8]

    Evaluating battery charging and swapping strategies in a robotic mobile fulfillment system , journal =

    Bipan Zou and Xianhao Xu and Yeming (Yale) Gong and René. Evaluating battery charging and swapping strategies in a robotic mobile fulfillment system , journal =. 2018 , issn =

    work page 2018

Show all 55 references
  1. [9]

    2020 , issn =

    Battery charge scheduling in long-life Autonomous Mobile Robots via multi-objective decision making under uncertainty , journal =. 2020 , issn =

    2020

  2. [10]

    2025 , issn =

    Deep Reinforcement Learning for dynamic order picking in Warehouse operations , journal =. 2025 , issn =

    2025

  3. [11]

    Reinforcement Learning for AMR Charging Decisions: The Impact of Reward and Action Space Design

    Bischoff, Janik and Rinciog, Alexandru and Meyer, Anne. Reinforcement Learning for AMR Charging Decisions: The Impact of Reward and Action Space Design. Learning and Intelligent Optimization. 2026

    2026

  4. [12]

    Real-Time Charging Scheduling of

    Lin, Chun-Cheng and Chen, Kun-Yang and Hsieh, Li-Tsung , journal=. Real-Time Charging Scheduling of. 2023 , volume=

    2023

  5. [13]

    Battery Management for Automated Warehouses via Deep Reinforcement Learning

    Deng, Yanchen and An, Bo and Qiu, Zongmin and Li, Liuxi and Wang, Yong and Xu, Yinghui. Battery Management for Automated Warehouses via Deep Reinforcement Learning. Distributed Artificial Intelligence. 2020

    2020

  6. [14]

    Neural Comput

    Zellner, Aaron and Dutta, Ayan and Kulbaka, Iliya and Sharma, Gokarna , title =. Neural Comput. Appl. , month = jun, pages =. 2023 , issue_date =

    2023

  7. [15]

    Proceedings of the 24th International Conference on Autonomous Agents and Multiagent Systems , pages =

    Pal, Aritra and Chauhan, Anandsingh and Baranwal, Mayank , title =. Proceedings of the 24th International Conference on Autonomous Agents and Multiagent Systems , pages =. 2025 , isbn =

    2025

  8. [16]

    2018 , editor =

    Rashid, Tabish and Samvelyan, Mikayel and Schroeder, Christian and Farquhar, Gregory and Foerster, Jakob and Whiteson, Shimon , booktitle =. 2018 , editor =

    2018

  9. [17]

    International Journal on Recent and Innovation Trends in Computing and Communication , volume =

    Jaiswal, Chandra , title =. International Journal on Recent and Innovation Trends in Computing and Communication , volume =

  10. [18]

    Solving the online batching problem using Deep Reinforcement Learning , journal =

    Bram Cals and Yingqian Zhang and Remco Dijkman and Claudy. Solving the online batching problem using Deep Reinforcement Learning , journal =. 2021 , issn =

    2021

  11. [19]

    2024 , eprint=

    Deep Reinforcement Learning for Picker Routing Problem in Warehousing , author=. 2024 , eprint=

    2024

  12. [20]

    2026 , howpublished =

    Čelik, Nejc and Škraba, Andrej , title =. 2026 , howpublished =

    2026

  13. [21]

    A Deep Reinforcement Learning Algorithm for Warehousing Multi-

    Shen, Guicheng and Ma, Ran and Tang, Zhong and Chang, Liangliang , booktitle=. A Deep Reinforcement Learning Algorithm for Warehousing Multi-. 2021 , volume=

    2021

  14. [22]

    Concurrency and Computation: Practice and Experience , volume =

    Dong, Tingting and Xue, Fei and Xiao, Chuangbai and Li, Juntao , title =. Concurrency and Computation: Practice and Experience , volume =

  15. [23]

    Technologies , VOLUME =

    Waubert de Puiseau, Constantin and Nanfack, Dimitri Tegomo and Tercan, Hasan and Löbbert-Plattfaut, Johannes and Meisen, Tobias , TITLE =. Technologies , VOLUME =. 2022 , NUMBER =

    2022

  16. [24]

    A Deep Reinforcement Learning Approach for Inventory Control under Stochastic Lead Time and Demand , year=

    Shakya, Manoj and Lee, Bu-Sung and Ng, Huey Yuen , booktitle=. A Deep Reinforcement Learning Approach for Inventory Control under Stochastic Lead Time and Demand , year=

  17. [25]

    Electronics , VOLUME =

    Zhang, Yang and He, Lili and Zheng, Junhong , TITLE =. Electronics , VOLUME =. 2025 , NUMBER =

    2025

  18. [26]

    2019 Winter Simulation Conference (WSC) , title=

    Li,. 2019 Winter Simulation Conference (WSC) , title=. 2019 , volume=

    2019

  19. [27]

    2020 , eprint=

    Duration-of-Stay Storage Assignment under Uncertainty , author=. 2020 , eprint=

    2020

  20. [28]

    Electrical Engineering in Japan , volume =

    Yoshida, Yasuhiro and Arai, Sachiyo and Kobayashi, Hiroyasu and Kondo, Keiichiro , title =. Electrical Engineering in Japan , volume =

  21. [29]

    Journal of Power Sources , volume =

    A review of. Journal of Power Sources , volume =. 2022 , issn =

    2022

  22. [30]

    Proceedings of the International Conference on Automated Planning and Scheduling , volume =

    Beeks, Martijn and Refaei Afshar, Reza and Zhang, Yingqian and Dijkman, Remco and van Dorst, Claudy and de Looijer, Stijn , title =. Proceedings of the International Conference on Automated Planning and Scheduling , volume =

  23. [31]

    Applied Sciences , VOLUME =

    Lee, HyeokSoo and Jeong, Jongpil , TITLE =. Applied Sciences , VOLUME =. 2021 , NUMBER =

    2021

  24. [32]

    Zhang, Lixiang and Yan, Yan and Hu, Yaoguang , title =. J. Intell. Manuf. , month = oct, pages =. 2023 , issue_date =

    2023

  25. [33]

    CAAI Transactions on Intelligence Technology , volume =

    Yang, Yang and Juntao, Li and Lingling, Peng , title =. CAAI Transactions on Intelligence Technology , volume =

  26. [34]

    and Li, Yuda and Martin, Xabier A

    Leon, Jonas F. and Li, Yuda and Martin, Xabier A. and Calvet, Laura and Panadero, Javier and Juan, Angel A. , TITLE =. Algorithms , VOLUME =. 2023 , NUMBER =

    2023

  27. [35]

    2019 , note =

    Design, Implementation and Evaluation of Reinforcement Learning for an Adaptive Order Dispatching in Job Shop Manufacturing Systems , journal =. 2019 , note =

    2019

  28. [36]

    A Deep Reinforcement Learning Based Approach for

    Guo, Xinde and Ren, Zhigang and Wu, Zongze and Lai, Jialun and Zeng, Deyu and Xie, Shengli , booktitle=. A Deep Reinforcement Learning Based Approach for. 2020 , volume=

    2020

  29. [37]

    Dispatching and Path Planning of Automated

    Zhang, Hongbin and Luo, Jiliang and Lin, Xinjie and Tan, Kaicheng and Pan, Chunrong , booktitle=. Dispatching and Path Planning of Automated. 2021 , volume=

    2021

  30. [38]

    2025 , volume=

    Mo, Wan and Yao, Weiran , booktitle=. 2025 , volume=

    2025

  31. [39]

    Sensors , VOLUME =

    Xiao, Qian and Pan, Tengteng and Wang, Kexin and Cui, Shuoming , TITLE =. Sensors , VOLUME =. 2025 , NUMBER =

    2025

  32. [40]

    An Improved

    Zhang, Jiatong and Lv, Yaqiong and Li, Yifan and Liu, Jialun , booktitle=. An Improved. 2022 , volume=

    2022

  33. [41]

    Path Planning of Autonomous Mobile Robot in Comprehensive Unknown Environment Using Deep Reinforcement Learning , year=

    Bai, Zekun and Pang, Hui and He, Zhaonian and Zhao, Bin and Wang, Tong , journal=. Path Planning of Autonomous Mobile Robot in Comprehensive Unknown Environment Using Deep Reinforcement Learning , year=

  34. [42]

    International Journal of Production Research , volume =

    Chen-Fu Chien and Yun-Siang Lin and Sheng-Kai Lin , title =. International Journal of Production Research , volume =. 2020 , publisher =

    2020

  35. [43]

    Deep Reinforcement Learning for the real-time inventory rack storage assignment and replenishment problem , journal =

    Sander Teck and T\'. Deep Reinforcement Learning for the real-time inventory rack storage assignment and replenishment problem , journal =. 2025 , issn =

    2025

  36. [44]

    International Journal of Production Research , volume =

    Francesco Stranieri and Fabio Stella and Chaaben Kouki , title =. International Journal of Production Research , volume =. 2024 , publisher =

    2024

  37. [45]

    2024 , issn =

    Information Fusion , volume =. 2024 , issn =

    2024

  38. [46]

    2022 IEEE 18th International Conference on Automation Science and Engineering (CASE) , pages =

    Lee, Min Seok and Jang, Young Jae , title =. 2022 IEEE 18th International Conference on Automation Science and Engineering (CASE) , pages =

    2022

  39. [47]

    Journal of Machine Learning Research , year =

    Antonin Raffin and Ashley Hill and Adam Gleave and Anssi Kanervisto and Maximilian Ernestus and Noah Dormann , title =. Journal of Machine Learning Research , year =

  40. [48]

    Proceedings of the 33rd International Conference on International Conference on Machine Learning - Volume 48 , pages =

    Wang, Ziyu and Schaul, Tom and Hessel, Matteo and Van Hasselt, Hado and Lanctot, Marc and De Freitas, Nando , title =. Proceedings of the 33rd International Conference on International Conference on Machine Learning - Volume 48 , pages =. 2016 , publisher =

    2016

  41. [49]

    2022 , volume=

    Mu, Yongjin and Li, Yanjie and Lin, Ke and Deng, Ki and Liu, Qi , booktitle=. 2022 , volume=

    2022

  42. [50]

    John Schulman and Philipp Moritz and Sergey Levine and Michael Jordan and Pieter Abbeel , year=. High-. 1506.02438 , archivePrefix=

    arXiv

  43. [51]

    Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining , pages =

    Akiba, Takuya and Sano, Shotaro and Yanase, Toshihiko and Ohta, Takeru and Koyama, Masanori , title =. Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining , pages =. 2019 , isbn =

    2019

  44. [52]

    Expert Systems with Applications , volume =

    Prediction of. Expert Systems with Applications , volume =. 2019 , issn =

    2019

  45. [53]

    and Lee, Su-In , title =

    Lundberg, Scott M. and Lee, Su-In , title =. Proceedings of the 31st International Conference on Neural Information Processing Systems , pages =. 2017 , isbn =

    2017

  46. [54]

    European Journal of Operational Research , volume=

    Reinforcement learning for multi-item retrieval in the puzzle-based storage system , author=. European Journal of Operational Research , volume=. 2023 , publisher=

    2023

  47. [55]

    2024 , publisher=

    Tian, Ran and Lu, Meng and Wang, Haopeng and Wang, Bo and Tang, Qingxia , journal=. 2024 , publisher=

    2024

Pith tools

Reviewed July 11, 2026 · model on record in the stance chip above.