Instantaneous Local Control Barrier Function: An Online Learning Approach for Collision Avoidance
This paper presents an integrated perception and control approach to accomplish safe autonomous navigation in unknown environments. This is achieved by numerical optimization with constraint learning for instantaneous local control barrier functions (IL-CBFs) and goal-driven control Lyapunov functions (GD-CLFs). In particular, the constraints reflecting safety and task requirements are first online learned from perceptual signals, wherein IL-CBFs are learned to characterize potential collisions, and GD-CLFs are constructed to reflect incrementally discovered subgoals. Then, the learned IL-CBFs are united with GD-CLFs in the context of a quadratic programming optimization, whose feasibility is improved by enlarging the shared control space. Numerical simulations are conducted to reveal the effectiveness of our proposed safe feedback control strategy that could drive the mobile robot to safely reach the destination incrementally in an uncertain environment.
PDF Abstract
