This paper presents a novel control strategy for mobile robots navigating in environments with occluded obstacles — situations where obstacles are hidden from the robot’s field of view but pose potential collision risks.
The Challenge: Robots relying on onboard sensors cannot detect obstacles behind occlusions, leading to collision risks and abrupt velocity changes that compromise motion stability.
Our Approach: We developed an occlusion-aware Consistent Model Predictive Control (CMPC) strategy that:
- Models “risk regions” — potential future locations of occluded obstacles
- Generates multiple trajectory branches simultaneously, each accounting for different risk scenarios
- Maintains a shared consensus segment across all trajectories to ensure smooth transitions and motion consistency
- Uses ADMM (Alternating Direction Method of Multipliers) to decompose the complex optimization into parallel sub-problems for real-time performance
Key Results:
- 38% reduction in lateral velocity variance compared to baseline approaches
- 51.7% lower peak lateral acceleration
- Real-time performance with ~40ms average solving time
- Successful validation in both simulation and real-world experiments on an Ackermann-steering robot platform
The method strikes an effective balance between safety (avoiding collisions with occluded obstacles) and performance (maintaining smooth, consistent motion), enabling robots to navigate confidently in cluttered, partially observable environments.