Design and test of the inertial navigation system for inspection robots in caged chicken coop

To address the challenges of obstacle perception and low navigation control accuracy caused by high cage density, weak echo signals from net-like structures, and narrow paths in caged chicken coops, an inertial navigation system is designed in this research based on multi-sensor perception for caged chicken coop inspection robots, which realizes autonomous navigation, obstacle avoidance, and teaching trajectory tracking of inspection robots in real working environments in caged chicken coops. Firstly, an ultrasonic sensor array was designed and optimized for the real and complex working environment in the caged chicken coop. The sliding window detection method was used to process the data sensed by multiple ultrasonic sensors, which improved the sensitivity and accuracy of the inspection robot in detecting obstacles on the side of the egg collector and feeder at the entrance of the aisle. Then, a pose correction algorithm based on infrared sensor data difference was proposed, which solved the cumulative error problem of inertial navigation by using the difference in obstacle distance between the two infrared sensors at the tail of the robot through the difference method, and achieved relative pose correction of the inspection robot in the caged chicken coop. Finally, a teaching trajectory tracking method based on position point and heading angle constraints was proposed. By collecting key point pose information on the robot's teaching trajectory and tracking it, the teaching trajectory tracking of the inspection robot in the caged chicken coop was achieved. The experimental results showed that the multi-sensor sensing and sliding window obstacle detection method proposed in this research has a good sensing effect on the hollow area on the side of the egg collector and the feeder at the entrance of the aisle. The obstacle distance sensing error was no more than 3 cm, and the inspection robot could enter the aisle entrance stably at a speed of 0.1-0.3 m/s and conducted inspections through the narrow area of the feeder. The differential pose correction algorithm using infrared sensors could correct poses under different angle errors, with a maximum lateral deviation of no more than 6 cm and a maximum angle deviation of no more than 3.89° after correction. This method effectively corrected the cumulative error of the inertial navigation of the inspection robot, and the inspection robot could smoothly enter the next aisle for operation after using the correction algorithm. The average deviation of the maximum position by using the teaching trajectory tracking method based on position point and heading angle constraints at different speeds was no more than 4.03 cm, the standard deviation of the maximum position was no more than 1.20 cm, the average deviation of the maximum angle was no more than 3.15°, and the standard deviation of the maximum angle was no more than 1.13°. The result showed that the obstacle avoidance navigation and teaching trajectory tracking method based on multi-sensor perception can achieve reliable obstacle avoidance and accurate autonomous navigation for robot inspection in the environment of the caged chicken coop. The proposed inertial navigation system has strong adaptability, good stability, and low cost, providing technical support for autonomous inspection and operation of caged chicken coops, as well as the intelligent development of the chicken breeding industry.