Centrala begrepp
The author presents a robust motion planning algorithm that learns obstacle uncertainties to improve safety and feasibility in dynamic environments.
Sammanfattning
The paper introduces an efficient motion-planning algorithm that learns obstacle uncertainties online to reduce conservatism. By predicting future motions of dynamic obstacles, the method ensures safe and feasible trajectories for robotic systems. The approach is validated through simulations and hardware experiments, demonstrating its effectiveness in real-world scenarios.
The research focuses on safe motion planning in dynamic environments, emphasizing the importance of predicting obstacle uncertainties accurately. By learning intended control sets of obstacles without prior assumptions, the proposed algorithm offers a less conservative solution compared to traditional methods. The study showcases the practical application of the method on a car-like mobile robot, highlighting its potential for real-world implementation.
Key contributions include proposing a novel approach to learning control sets efficiently and designing a robust predictive motion planner for collision avoidance. The method's performance is evaluated through simulations and hardware experiments, showcasing its ability to handle uncertain surroundings effectively. Overall, the paper provides valuable insights into improving motion planning algorithms for robotic systems.
Statistik
The worst-case distance between EV and SV with DMPC is 0.03 m.
The minimum EV-SV distance with the proposed method is 0.14 m.
The cost function value ranges from 2000 to 2331.
Citat
"The worst-case characterization gives a conservative uncertainty prediction."
"Learning intended control sets enables better motion prediction."
"The proposed method reduces conservatism while maintaining safety."