An adaptive variable-structure duct
robot motion control method based on
environmental sensing. A duct
robot is a dual-swing-arm tracked mechanism, and onboard sensing devices include a depth camera, an IMU, a
wheel speed sensor and a TOF module. The method comprises: a depth camera capturing an image of the interior of a duct to perform
scenario recognition, segmenting planar and curved surfaces in the image on the basis of
scenario recognition results, and extracting edge lines of the duct; using the edge lines to calculate the width of a straight duct, the curvature of a curved duct, the angle of a slope and the height of a step, in order to determine the passability of a
robot; designing a path planner and a swing arm planner to generate a reference trajectory and a swing arm angle sequence of the robot, performing
smoothing processing, and inputting the smoothed reference trajectory and swing arm angle sequence into an MPC
motion controller; using data from the depth camera, an IMU and a
wheel speed sensor to estimate the position and state of the robot by means of an ESKF
algorithm, and inputting
estimation results and a collision warning
signal from a TOF module into the MPC
motion controller in real time; and finally, outputting control signals for a locomotion
electric motor and a swing arm
electric motor. Thus, the adaptive variable-structure and autonomous motion of a robot in different scenarios is realized.