Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
169 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Roadmaps with Gaps over Controllers: Achieving Efficiency in Planning under Dynamics (2310.03239v5)

Published 5 Oct 2023 in cs.RO

Abstract: This paper aims to improve the computational efficiency of motion planning for mobile robots with non-trivial dynamics through the use of learned controllers. Offline, a system-specific controller is first trained in an empty environment. Then, for the target environment, the approach constructs a data structure, a "Roadmap with Gaps," to approximately learn how to solve planning queries using the learned controller. The roadmap nodes correspond to local regions. Edges correspond to applications of the learned controller that approximately connect these regions. Gaps arise as the controller does not perfectly connect pairs of individual states along edges. Online, given a query, a tree sampling-based motion planner uses the roadmap so that the tree's expansion is informed towards the goal region. The tree expansion selects local subgoals given a wavefront on the roadmap that guides towards the goal. When the controller cannot reach a subgoal region, the planner resorts to random exploration to maintain probabilistic completeness and asymptotic optimality. The accompanying experimental evaluation shows that the approach significantly improves the computational efficiency of motion planning on various benchmarks, including physics-based vehicular models on uneven and varying friction terrains as well as a quadrotor under air pressure effects.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (36)
  1. S. M. LaValle and J. J. Kuffner Jr, “Randomized kinodynamic planning,” IJRR, vol. 20, no. 5, pp. 378–400, 2001.
  2. Y. Li, Z. Littlefield, and K. E. Bekris, “Asymptotically optimal sampling-based kinodynamic planning,” IJRR, vol. 35, no. 5, pp. 528–564, 2016.
  3. K. Hauser and Y. Zhou, “Asymptotically optimal planning by feasible kinodynamic planning in a state–cost space,” T-RO, vol. 32, no. 6, pp. 1431–1443, 2016.
  4. Z. Littlefield and K. E. Bekris, “Efficient and asymptotically optimal kinodynamic motion planning via dominance-informed regions,” in IROS, 2018.
  5. M. Kleinbort, E. Granados, K. Solovey, R. Bonalli, K. E. Bekris, and D. Halperin, “Refined analysis of asymptotically-optimal kinodynamic planning in the state-cost space,” in ICRA, 2020.
  6. T. McMahon, A. Sivaramakrishnan, E. Granados, and K. E. Bekris, “A survey on the integration of machine learning with sampling-based motion planning,” Foundations and Trends® in Robotics, vol. 9, no. 4, pp. 266–327, 2022.
  7. A. Faust, O. Ramírez, M. Fiser, K. Oslund, A. Francis, J. O. Davidson, and L. Tapia, “PRM-RL: Long-range robotic navigation tasks by combining reinforcement learning and sampling-based planning,” ICRA, 2018.
  8. H.-T. L. Chiang, J. Hsu, M. Fiser, L. Tapia, and A. Faust, “RL-RRT: Kinodynamic motion planning via learning reachability estimators from RL policies,” RA-L, vol. 4, no. 4, pp. 4298–4305, 2019.
  9. Y. Li, R. Cui, Z. Li, and D. Xu, “Neural network approximation based near-optimal motion planning with kinodynamic constraints using RRT,” IEEE Transactions on Industrial Electronics, vol. 65, pp. 8718–8729, 2018.
  10. G. P. Kontoudis and K. G. Vamvoudakis, “Kinodynamic motion planning with continuous-time q-learning: An online, model-free, and safe navigation framework,” IEEE transactions on neural networks and learning systems, vol. 30, no. 12, pp. 3803–3817, 2019.
  11. J. J. Johnson, L. Li, F. Liu, A. H. Qureshi, and M. C. Yip, “Dynamically constrained motion planning networks for non-holonomic robots,” in IROS, 2020, pp. 6937–6943.
  12. L. Li, Y. Miao, A. H. Qureshi, and M. C. Yip, “Mpc-mpnet: Model-predictive motion planning networks for fast, near-optimal planning under kinodynamic constraints,” RA-L, vol. 6, no. 3, 2021.
  13. Z. Xu, B. Liu, X. Xiao, A. Nair, and P. Stone, “Benchmarking reinforcement learning techniques for autonomous navigation,” in ICRA, 2023.
  14. B. Eysenbach, R. R. Salakhutdinov, and S. Levine, “Search on the replay buffer: Bridging planning and reinforcement learning,” NeurIPS, 2019.
  15. S. Emmons, A. Jain, M. Laskin, T. Kurutach, P. Abbeel, and D. Pathak, “Sparse graphical memory for robust planning,” NeurIPS, 2020.
  16. A. Bagaria and G. Konidaris, “Option discovery using deep skill chaining,” in ICLR, 2019.
  17. A. Bagaria, J. K. Senthil, M. Slivinski, and G. Konidaris, “Robustly learning composable options in deep reinforcement learning.” in IJCAI, 2021.
  18. A. Bagaria, J. K. Senthil, and G. Konidaris, “Skill discovery for exploration and planning using deep skill graphs,” in International Conference on Machine Learning.   PMLR, 2021, pp. 521–531.
  19. R. Strudel, R. Garcia, J. Carpentier, J.-P. Laumond, I. Laptev, and C. Schmid, “Learning obstacle representations for neural motion planning,” in CoRL, 2020.
  20. A. Sivaramakrishnan, E. Granados, S. Karten, T. McMahon, and K. E. Bekris, “Improving kinodynamic planners for vehicular navigation with learned goal-reaching controllers,” in IROS, 2021.
  21. T. McMahon, A. Sivaramakrishnan, K. Kedia, E. Granados, and K. E. Bekris, “Terrain-aware learned controllers for sampling-based kinodynamic planning over physically simulated terrains,” in IROS, 2022.
  22. D. Le and E. Plaku, “Guiding sampling-based tree search for motion planning with dynamics via probabilistic roadmap abstractions,” in IROS, 2014.
  23. M. G. Westbrook and W. Ruml, “Anytime kinodynamic motion planning using region-guided search,” in IROS, 2020.
  24. R. Shome and L. E. Kavraki, “Asymptotically optimal kinodynamic planning using bundles of edges,” in ICRA, 2021.
  25. E. Todorov, T. Erez, and Y. Tassa, “Mujoco: A physics engine for model-based control,” in IROS, 2012.
  26. T. Haarnoja, A. Zhou, P. Abbeel, and S. Levine, “Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor,” in ICML, 2018.
  27. M. Andrychowicz, D. Crow, A. Ray, J. Schneider, R. H. Fong, P. Welinder, B. McGrew, J. Tobin, P. Abbeel, and W. Zaremba, “Hindsight experience replay,” in NeurIPS, 2017.
  28. S. S. Srinivasa, P. Lancaster, J. Michalove, M. Schmittle, C. Summers, M. Rockett, J. R. Smith, S. Chouhury, C. Mavrogiannis, and F. Sadeghi, “MuSHR: A low-cost, open-source robotic racecar for education and research,” CoRR, vol. abs/1908.08031, 2019.
  29. E. Granados, A. Sivaramakrishnan, T. McMahon, Z. Littlefield, and K. E. Bekris, “Machine learning for kinodynamic planning (ml4kp),” https://github.com/PRX-Kinodynamic/ML4KP, 2021–2022.
  30. W. Hönig, J. Ortiz-Haro, and M. Toussaint, “db-a*: Discontinuity-bounded search for kinodynamic mobile robot motion planning,” in IROS, 2022.
  31. L. Kästner, T. Bhuiyan, T. A. Le, E. Treis, J. Cox, B. Meinardus, J. Kmiecik, R. Carstens, D. Pichel, B. Fatloun, et al., “Arena-bench: A benchmarking suite for obstacle avoidance approaches in highly dynamic environments,” RA-L, vol. 7, no. 4, 2022.
  32. E. Heiden, L. Palmieri, L. Bruns, K. O. Arras, G. S. Sukhatme, and S. Koenig, “Bench-mr: A motion planning benchmark for wheeled mobile robots,” RA-L, vol. 6, no. 3, 2021.
  33. J. Fu, A. Kumar, O. Nachum, G. Tucker, and S. Levine, “D4rl: Datasets for deep data-driven reinforcement learning,” 2020.
  34. K. Stachowicz, A. Bhorkar, D. Shah, I. Kostrikov, and S. Levine, “FastRLAP: A System for Learning High-Speed Driving via Deep RL and Autonomous Practicing,” 2023. [Online]. Available: TODO
  35. A. Dobson and K. E. Bekris, “Sparse roadmap spanners for asymptotically near-optimal motion planning,” IJRR, vol. 33, no. 1, pp. 18–47, 2014.
  36. S. Kiesel, T. Gu, and W. Ruml, “An effort bias for sampling-based motion planning,” in IROS, 2017.
Citations (2)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now