Research on inertial navigation system for inspection robot in caged chicken coop

The structure of caged chicken coop is complicated and there are many obstacles, which makes it difficult to detect obstacles of inspection robot. At the same time, the inspection routes between different chicken coops vary greatly, resulting in high cost of visual navigation and lidar navigation, complex path planning, and high application cost of inspection operations. To solve the above problems, this study designed an inertial navigation for caged chicken coop inspection robot, and realized the autonomous obstacle avoidance navigation and low-cost flexible and fast operation application of the inspection robot in the complex environment of the caged chicken coop. Firstly, aiming at the complex environment in the cage chicken coop, the ultrasonic array sensors were designed to process the ultrasonic data through multi-sensor sensing and sliding window detection methods, which improved the inspection robot's ability and accuracy of perceiving obstacles on the side of the egg collector and the feeder at the entrance of the aisle. Then, the pose correction algorithm based on sensor difference method was proposed. The accumulated error of inertial navigation was solved by using the distance difference between the infrared sensors, and the pose correction of the inspection robot in the cage chicken coop was realized. Finally, an inspection path planning and reproduction method based on teaching learning and track tracking was proposed. The key pose points were used to record the action of teaching learning, and the pose and heading angle constraints were used to achieve path tracking, which realized the rapid inspection path planning and operation application of the inspection robot in the cage chicken coop. It solved the problems of complicated path planning and difficult operation of inspection robots. The test results showed that the multi-sensor sensing and sliding window obstacle detection method proposed in this study has a good perception effect on the hollow-out on the side of the egg collector at the entrance of the aisle and the feeder, and the perception error of the obstacle distance was less than 3 cm, which can realize the stable entry into the aisle at the speed of 0.1～0.3 m/s and the inspection through the narrow place of the feeder. The pose correction algorithm using the sensor difference method can correct the pose under different angle errors, and the maximum lateral deviation after correction was less than 6 cm and the maximum angle deviation was less than 3.89°, which met the accuracy requirements of the robot entering the next aisle and effectively corrected the accumulated error of the inertial navigation of the inspection robot. The average deviation of the maximum position reproduced by the inspection route planned through teaching learning at different speeds was less than 4.03 cm, the standard deviation of the maximum position was less than 1.20 cm, the average deviation of the maximum angle was less than 3.15°, and the standard deviation of the maximum angle was less than 1.13°. The route planning method for caged chicken coop inspection robot based on teaching and learning and path tracking proposed in this study can meet the accuracy requirements of the robot in the caged chicken coop, and can realize the robot to the inspection path accurately and stably. The navigation system can reduce the application cost of inspection robot in caged chicken coop through low cost, simple operation and good stability of inertial navigation system, which provides technical support for automatic inspection of caged chicken coop and intelligent development of caged chicken breeding industry.