Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
140 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
46 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

Conservative Linear Envelopes for Nonlinear, High-Dimensional, Hamilton-Jacobi Reachability (2403.14184v3)

Published 21 Mar 2024 in eess.SY, cs.SY, and math.OC

Abstract: Hamilton-Jacobi reachability (HJR) provides a value function that encodes the set of states from which a system with bounded control inputs can reach or avoid a target despite any bounded disturbance, and the corresponding robust, optimal control policy. Though powerful, traditional methods for HJR rely on dynamic programming (DP) and suffer from exponential computation growth with respect to state dimension. The recently favored Hopf formula mitigates this curse of dimensionality'' by providing an efficient and space-parallelizable approach for solving the reachability problem. However, the Hopf formula can only be applied to linear time-varying systems. To overcome this limitation, we show that the error between a nonlinear system and a linear model can be transformed into an adversarial bounded artificial disturbance. One may then solve the dimension-robust generalized Hopf formula for a linear game with thisantagonistic error" to perform guaranteed conservative reachability analysis and control synthesis of nonlinear systems; this can be done for problem formulations in which no other HJR method is both computationally feasible and guaranteed. In addition, we offer several technical methods for reducing conservativeness in the analysis. We demonstrate the effectiveness of our results through one illustrative example (the controlled Van der Pol system) that can be compared to standard DP, and one higher-dimensional 15D example (a 5-agent pursuit-evasion game with Dubins cars).

Definition Search Book Streamline Icon: https://streamlinehq.com
References (28)
  1. X. Chen, E. Abraham, and S. Sankaranarayanan, “Taylor model flowpipe construction for non-linear hybrid systems,” in 2012 IEEE 33rd Real-Time Systems Symposium.   IEEE, 2012, pp. 183–192.
  2. S. Bogomolov, M. Forets, G. Frehse, K. Potomkin, and C. Schilling, “Juliareach: a toolbox for set-based reachability,” in Proceedings of the 22nd ACM International Conference on Hybrid Systems: Computation and Control, 2019, pp. 39–44.
  3. I. M. Mitchell, A. M. Bayen, and C. J. Tomlin, “A time-dependent hamilton-jacobi formulation of reachable sets for continuous dynamic games,” IEEE Transactions on automatic control, vol. 50, no. 7, pp. 947–957, 2005.
  4. S. Bansal, M. Chen, S. Herbert, and C. J. Tomlin, “Hamilton-Jacobi reachability: A brief overview and recent advances,” in 2017 IEEE 56th Annual Conference on Decision and Control (CDC).   IEEE, 2017, pp. 2242–2253.
  5. M. R. Kirchner, R. Mar, G. Hewer, J. Darbon, S. Osher, and Y. T. Chow, “Time-optimal collaborative guidance using the generalized Hopf formula,” IEEE Control Systems Letters, vol. 2, no. 2, pp. 201–206, apr 2018. [Online]. Available: https://doi.org/10.11092Flcsys.2017.2785357
  6. S. L. Herbert, M. Chen, S. Han, S. Bansal, J. F. Fisac, and C. J. Tomlin, “Fastrack: A modular framework for fast and guaranteed safe motion planning,” in 2017 IEEE 56th Annual Conference on Decision and Control (CDC).   IEEE, 2017, pp. 1517–1522.
  7. S. Bansal and C. J. Tomlin, “Deepreach: A deep learning approach to high-dimensional reachability,” in 2021 IEEE International Conference on Robotics and Automation (ICRA).   IEEE, 2021, pp. 1817–1824.
  8. W. Sharpless, N. Shinde, M. Kim, Y. T. Chow, and S. Herbert, “Koopman-hopf hamilton-jacobi reachability and control,” arXiv preprint arXiv:2303.11590, 2003.
  9. M. Althoff, O. Stursberg, and M. Buss, “Reachability analysis of nonlinear systems with uncertain parameters using conservative linearization,” in 2008 47th IEEE Conference on Decision and Control.   IEEE, 2008, pp. 4042–4048.
  10. J. K. Scott and P. I. Barton, “Bounds on the reachable sets of nonlinear control systems,” Automatica, vol. 49, no. 1, pp. 93–100, 2013.
  11. M. Althoff and B. H. Krogh, “Zonotope bundles for the efficient computation of reachable sets,” in 2011 50th IEEE conference on decision and control and European control conference.   IEEE, 2011, pp. 6814–6821.
  12. I. Rublev, “Generalized Hopf formulas for the nonautonomous Hamilton-Jacobi equation,” Computational Mathematics and Modeling, vol. 11, no. 4, pp. 391–400, 2000.
  13. Y. T. Chow, J. Darbon, S. Osher, and W. Yin, “Algorithm for overcoming the curse of dimensionality for certain non-convex hamilton–jacobi equations, projections and differential games,” Annals of Mathematical Sciences and Applications, vol. 3, no. 2, pp. 369–403, 2018.
  14. A. B. Kurzhanski, “Dynamics and control of trajectory tubes. theory and computation,” in 2014 20th International Workshop on Beam Dynamics and Optimization (BDO).   IEEE, 2014, pp. 1–1.
  15. J. Darbon and S. Osher, “Algorithms for overcoming the curse of dimensionality for certain hamilton–jacobi equations arising in control theory and elsewhere,” Research in the Mathematical Sciences, vol. 3, no. 1, p. 19, 2016.
  16. Y. T. Chow, J. Darbon, S. Osher, and W. Yin, “Algorithm for overcoming the curse of dimensionality for time-dependent non-convex hamilton–jacobi equations arising from optimal control and differential games problems,” Journal of Scientific Computing, vol. 73, pp. 617–643, 2017.
  17. D. Lee and C. J. Tomlin, “Iterative method using the generalized Hopf formula: Avoiding spatial discretization for computing solutions of Hamilton-Jacobi equations for nonlinear systems,” in 2019 IEEE 58th Conference on Decision and Control (CDC).   IEEE, 2019, pp. 1486–1493.
  18. A. F. Filippov, “Differential equations with discontinuous right-hand side,” Matematicheskii sbornik, vol. 93, no. 1, pp. 99–128, 1960.
  19. L. C. Evans and P. E. Souganidis, “Differential games and representation formulas for solutions of Hamilton-Jacobi-isaacs equations,” Indiana University mathematics journal, vol. 33, no. 5, pp. 773–797, 1984.
  20. E. Hopf, “Generalized solutions of non-linear equations of first order,” Journal of Mathematics and Mechanics, vol. 14, no. 6, pp. 951–973, 1965.
  21. M. Bardi and L. C. Evans, “On hopf’s formulas for solutions of hamilton-jacobi equations,” Nonlinear Analysis: Theory, Methods & Applications, vol. 8, no. 11, pp. 1373–1381, 1984.
  22. P.-L. Lions and J.-C. Rochet, “Hopf formula and multitime Hamilton-Jacobi equations,” Proceedings of the American Mathematical Society, vol. 96, no. 1, pp. 79–84, 1986.
  23. Y. T. Chow, J. Darbon, S. Osher, and W. Yin, “Algorithm for overcoming the curse of dimensionality for state-dependent Hamilton-Jacobi equations,” Journal of Computational Physics, vol. 387, pp. 376–409, 2019.
  24. A. I. Subbotin, “Minimax solutions of first-order partial differential equations,” Russian Mathematical Surveys, vol. 51, no. 2, p. 283, 1996.
  25. Q. Wei, “Viscosity solution of the hamilton–jacobi equation by a limiting minimax method,” Nonlinearity, vol. 27, no. 1, p. 17, 2013.
  26. Y. Wang, W. Yin, and J. Zeng, “Global convergence of admm in nonconvex nonsmooth optimization,” Journal of Scientific Computing, vol. 78, pp. 29–63, 2019.
  27. L. C. Evans, “Envelopes and nonconvex hamilton–jacobi equations,” Calculus of Variations and Partial Differential Equations, vol. 50, no. 1-2, pp. 257–282, 2014.
  28. A. Bemporad, F. Borrelli, M. Morari et al., “Model predictive control based on linear programming~ the explicit solution,” IEEE transactions on automatic control, vol. 47, no. 12, pp. 1974–1985, 2002.
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