Probabilistic Safety-Assured Adaptive Merging Control for Autonomous Vehicles

Adaptive Safe Merging Control for Heterogeneous Autonomous Vehicles using Parametric Control Barrier Functions

“…Motivated by these considerations, we focus on the learning and safe design for interaction of heterogeneous autonomous vehicles in ramp merging scenario. This paper extends our previous work on CBF-based safety-assured merging control in [5] and presents the following contributions: 1) We propose the novel idea of Parametric-CBF, a variant of traditional CBF that gives a richer behavior description and preserves the property of rendering a forward invariant safe set for the robots; 2) We present a novel safe adaptive merging algorithm that integrates the safe behavior prediction of heterogeneous robots and safe control for the ego robot using the learned parameters of the Parametric CBF, yielding improved task efficiency for the ego robot with safety guarantee. 3) We demonstrate the effectiveness of the Parametric-CBF based behavior predication and safe control through experimental results in the task of ramp merging in the autonomous driving domain; Our mechanism enables the robot to model the behavior of other entities first and take appropriate actions accordingly, which makes it generally applicable to other robotics applications.…”

“…In this work, the system dynamics of a vehicle can be described by the same double integrators as in [5], since acceleration plays a key role in the safety considerations:…”

“…Among the model-based safe control approaches, Control Barrier Function (CBF) is a popular choice due to its forward invariant property in enforcing persistent safety. In recent progress on safe design for autonomous robots using CBF [1]- [5], surrounding robots are often assumed to be fully cooperative or passively moving with constant velocity in the collision avoidance scenario. However, in a more realistic setting when robots operate in an unknown environment and interact with other autonomous systems, e.g.…”

Provable Probabilistic Safety and Feasibility-Assured Control for Autonomous Vehicles using Exponential Control Barrier Functions

“…Moreover, for optimization-based safety-critical controllers that use CBF-based constraints, solution feasibility is a common problem. [8] mentions that the solution feasibility can be guaranteed by assuming that, in the worst case, making all robots decelerate to zero velocity immediately at the next time step can always prevent collision, and therefore the feasible solution space will always be non-empty. However, a more principled scheme with explicit theoretical grounding is desired to automatically decide whether the vehicle needs to perform full braking before it is too late.…”

“…As a modern control approach, Control Barrier Functions (CBF) [6], [7] offers a more admissive control space based on the idea of set forward invariance, which is formally provable. Therefore, it is seeing increased usage in safe control [8]- [11]. When it comes to complex systems, Exponential CBF (eCBF) demonstrates its usefulness in high-relative-degree safety-critical control [6], [9].…”