Singularity-free high-precision docking control for 2-SD nursing robots based on multimodal perception

Abstract

To address maneuverability constraints during contactless bed-chair transfers for disabled patients, this paper utilizesa2-SD (steer-drive) composite chassis for nursing robots. A multi-modal autonomous docking control system is proposed to resolve kinematic singularities during orthogonal steering and near-field optical perception limitations. First, a time-domain "zero-velocity interlock" mechanism isolates orthogonal transitions, preventing singularity-induced tire slip and mechanical tearing. Second, for lateral translation, a bidirectional nonlinear control combining step damping and friction feedforward mitigates ultra-low-speed motor stalls. Crucially, to safely accommodate mechanical reconfiguration, the system halts at a 30 cm boundary to transition the patient into a lying posture and deploy a lateral V-shaped transfer board. Subsequently, to navigate the inevitable 10 cm LiDAR blind spot during the final approach, the system employs an odometry pulse dead-reckoning strategy. Real-vehicle tests demonstrate that under various initial poses, the terminal docking error is consistently bounded within 2.1 mm. This high-precision control ensures a seamless physical bridge for the transfer board while strictly avoiding mechanical collisions with the bed frame, providing a highly reliable engineering solution for restricted clinical spaces.