Model-Based Probabilistic Collision Detection in Autonomous Driving

Abstract

Abstract-Safety of planned paths of autonomous cars with respect to the movement of other traffic participants is considered. Thereto, the stochastic occupancy of the road by other vehicles is predicted. The prediction considers uncertainties originating from the measurements and the possible behaviors of other traffic participants. In addition, the interaction of traffic participants as well as the limitation of driving maneuvers due to the road geometry is considered. The result of the presented approach is the probability of a crash for a specific trajectory of the autonomous car. The presented approach is efficient as most intensive computations are performed offline, resulting in a lean online algorithm for real-time application.

Bayesian environment representation, prediction, and criticality assessment for driver assistance systems

“…Parametric descriptions are to date only common for road course models or road boundary estimators and not for general free space estimators. 5 Common road course models are, for example, circular models [107], cubic clothoid approximations [131], clothoids [65], or, recently, spline functions [9,156,243,264,265] for greater flexibility.…”

“…The set of environment sensor measurements is generated by a stereo camera and an automotive long-range radar sensor with individual appropriate inverse sensor models in this work, see [91]. 5 The set of ego vehicle poses is supposed to be known precisely due to the ego vehicle's proprioceptive sensors. Each of the three parts is described in detail in the following subsections.…”

“…• Methods that fuse dynamic motion models with behavior and environment descriptions [5,45,75,82,88,144,211,236].…”

“…In [5], a similar approach is presented, which is based on the determination of stochastic reachable sets of all other, interacting traffic participants under the assumption that these follow specific paths along a known road network with a certain accuracy. The ego vehicle's future path is considered available as the system is designed for autonomous vehicles with a known planned future trajectory and not for ADAS.…”

Motion planning under uncertainty for on-road autonomous driving

“…If the reachable set of an autonomous vehicle has no overlap with the reachable set of obstacles, the planned trajectory is marked as safe. In [12], the possible behaviors of traffic participants are also predicted and the reachable sets of traffic participants are computed based on a Markov chain. The limitation of this safety verification method is that it relies on the planned trajectory rather than being embedded in the planner.…”