Mobile-Master-Vehicle-Based LiDAR Perception and Centralized Control of Sensor-Light Slave Vehicles.

This study presents an asymmetric cooperative perception-and-control architecture in which a sensor-rich mobile master vehicle performs external perception, state estimation, motion planning, and path-tracking control for sensor-light slave vehicles. The proposed vehicle-model-aided LiDAR tracking (VMALT) method combines ego-motion-compensated LiDAR measurements with transmitted speed and steering commands through a kinematic bicycle model. In physical-vehicle experiments, VMALT reduced the slave-position RMSE from 0.36 m with a LiDAR-only constant-velocity Kalman filter to 0.32 m, corresponding to an 11.1% improvement. The estimated state was subsequently used for closed-loop speed and path-tracking control of the physical slave vehicle. Scalability was evaluated using a one-master-two-slave configuration over three separate runs comprising circular and linear paths and geometrically identified line-of-sight-overlap candidates. Concurrent two-target LiDAR availability ranged from 96.465% to 99.934%, with a maximum interior observation gap of 0.10 s. The VMALT trajectories expressed in the GPS coordinate frame followed the position and direction of both slave vehicles, with heading RMSEs ranging from 1.835∘ to 4.609∘. In an LTE-tethering communication test, all 2400 UDP packets were returned, with median and 99th-percentile round-trip times of 4.654 and 8.220 ms, respectively. These results demonstrate the feasibility of centralized perception and control for multiple sensor-light vehicles under the evaluated low-speed operating conditions.