Set-Based Prediction of Traffic Participants Considering Occlusions and Traffic Rules

Abstract

Provably safe motion planning for automated road vehicles must ensure that planned motions do not result in a collision with other traffic participants. This is a major challenge in autonomous driving, since the future behavior of other traffic participants is not known and since traffic participants are often hidden due to occlusions. In this work, we propose a formal setbased prediction that contains all acceptable future behaviors of both detected and potentially hidden traffic participants. Based on formalized traffic rules and nondeterministic motion models, we perform reachability analysis to predict the set of possible occupancies and velocities of vehicles, pedestrians, and cyclists. Real-world experiments with a test vehicle in various traffic situations demonstrate the applicability and real-time capability of our over-approximative prediction for both online verification and fail-safe trajectory planning. Even in congested, complex traffic scenarios, our forecasting approach enables self-driving vehicles to never cause accidents.

Risk-aware Motion Planning for Autonomous Vehicles with Safety Specifications

“…Set-based reachability analysis allows one to compute the set of states that a system is able to reach over time considering an initial set of states and all admissible control inputs [13]. This technique can be used to determine if the AV is able to reach any unsafe set of states [14], [15] or to rigorously predict the possible occupied space of other traffic participants [16]- [18]. Reachability analysis is well suited to verify the safety of the AV during its operation as shown in [19]- [22].…”

Safe Occlusion-Aware Autonomous Driving via Game-Theoretic Active Perception

“…Recent works [3][4][5][6]18] tackle the risk assessment problem under occlusion by introducing "phantom" objects beyond the field of view and predicting their future states through the forward reachability analysis. Yu et al [5,6] apply a sampling method to represent potentially occluded objects' states as particles.…”

“…where υõ is a non-anticipative strategy for the pursuer [34]; this is equivalent to both players performing state feedback 3 .…”

“…For autonomous vehicles to operate safely, risks outside their field of view must be evaluated while planning for future actions. Previous works [3][4][5][6] utilize forward reachability analysis to over-approximate potential occluded objects' Fig. 1: The autonomous vehicle (evader) verifies the safety of two potential trajectoriesx e andx e against collisions with a hidden agent (pursuer) outside the current field of view Z t0 .…”

“…For example, we may require our ego vehicle, but not the hidden pursuer, to avoid other potentially existing objects 3. The equivalence is due to Isaacs' condition; we direct interested readers to[9, Ch.…”