Learning Reachability-Based Safety Representations for Real-Time Robot Navigation in Unknown and Dynamic Environments

Key points

• Hamilton--Jacobi (HJ) reachability analysis captures this in a value function whose sign separates states that admit a safe continuation from those where collision is unavoidable, but computing it online for every newly observed environment is too expensive.1
• This dissertation trains neural implicit representations offline on reachability solutions to give optimization-based controllers this information in real time.1
• The first two of five methods use the learned function as a terminal constraint in model predictive control, for static local environments and predicted obstacle motion.1
• For control-affine dynamics with polytopic inputs, the HJ Hamiltonian is maximized at a vertex, so a parallelizable tree search over vertex inputs can generate training labels.1

Sources (1)

• [1]Learning Reachability-Based Safety Representations for Real-Time Robot Navigation in Unknown and Dynamic Environments
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 12, 10:10 AM

  Safe robot navigation in unknown, dynamic environments requires more than geometric collision checking: a robot still far from an obstacle may be unable to avoid it because of its speed, orientation, or the obstacle's motion.

  Hamilton--Jacobi (HJ) reachability analysis captures this in a value function whose sign separates states that admit a safe continuation from those where collision is unavoidable, but computing it online for every newly observed environment is too expensive.

Extractive summary: sentences quoted from the sources.