Risk-aware model improves safe trajectory planning for self-driving cars
Risk-Aware World Modeling with Flow-Guided Occupancy Evolution for Selective Trajectory Planning in Automated Driving
Artificial IntelligenceEmerging TechnologiesMachine LearningRobotics
Summary
Planning safe paths for self-driving cars means predicting where other cars and pedestrians might be and what risks might happen soon. The authors introduce RiskWorld, a system that predicts how the positions of other road users will change and checks if the current driving plan is still safe. It only changes the planned path if the predicted risks increase beyond a limit and there is a better alternative path. Tests show RiskWorld reduces collisions and keeps errors low while running efficiently.
What this means in practice
- •For autonomous vehicle developers: Improve real-time decision making by selectively updating planned trajectories only when predicted risks rise, enhancing safety and efficiency of self-driving cars.
- •For robotics software engineers: Incorporate risk-aware occupancy forecasting into navigation systems for mobile robots to better anticipate dynamic obstacles and avoid collisions.
Authors
Rongxiang Zeng, Linsen Cai, Jiafu Zhang, Yijie Zhong, Yide Tao, Shuai Wang, Nan Zheng, Hai L. Vu, Alvaro Garcia Hernandez, Yongqi Dong
Abstract
Safe motion planning in automated driving requires anticipating evolving traffic risks and deciding when to revise the current planned trajectory. We introduce RiskWorld, a risk-aware world modeling framework for shared occupancy forecasting and selective trajectory replacement. Spatial risk fields and temporal actor context are fused with visual bird's-eye-view features. Flow-guided evolution transports occupancy and scene features, while signed residuals correct occupancy after transport. One forecast is generated per planning step and reused across candidates. Each candidate is compared with a current-state persistence reference, yielding a nonnegative collision-score correction. The trajectory selected by current-world evaluation serves as the planning anchor and is replaced only when additional predicted risk triggers intervention and an alternative satisfies component-wise constraints on predicted risk and trajectory error. Candidate geometries remain unchanged. We evaluate RiskWorld for open-loop planning on nuScenes using camera features, annotation-derived current and historical actor states, and dataset-provided map context. RiskWorld achieves the lowest collision rate at a long evaluation horizon of 3 s, and the second-best average L2 error among various state-of-the-art baselines, while running at 11.5 FPS on a single NVIDIA RTX 4090 with 90.81 M parameters. Within-setting ablations show that RiskWorld achieves lower collision rates than the current-state rescoring baseline, while forecast reuse enables additional candidates to be evaluated at low marginal computational cost.