Embedded Files
In the field of robotics, an autonomous navigation system is crucial for robots to move independently. The "Return to Home Challenge" aimed to develop autonomous navigation in GPS-denied indoor environments. Initially using an Extended Kalman Filter (EKF) faced sensor issues, leading to a switch to a 2D LiDAR-based SLAM algorithm. The revised plan maintained a PID controller for homing, incorporating waypoints for better accuracy. The system, using SLAM positional data, smoothly navigated and adapted to environmental changes during testing.
FIGure 1. SLAM MAP OF TEsts
FIGure 1. SLAM MAP OF TEsts
Figure 2. SLAM MAP PATH 1
Figure 2. SLAM MAP PATH 1
Figure 3. SLAM MAP PATH 2
Figure 3. SLAM MAP PATH 2
TESTING OF THE RETURN TO HOME SOLUTION:
IMG_6249 (1).MOVDuring the Return to Home Challenge demonstration, Lidar maps (Figures 2 and 3) showed unexpected deviations from pre-demo tests, mainly due to environmental noise and hardware issues. The robot's performance, anticipated to be similar to Figure 1, differed significantly. In both runs, we faced challenges, including readiness issues, connection delays, port configuration shifts, Matlab logouts, and an undercharged battery causing a shutdown. Run two also suffered from mapping errors, excessive noise, and faulty sensors, leading to failure in the autonomous homing sequence.
To enhance the code base, suggested improvements include setting parameters for loop closures, tuning scan numbers for open environments, and implementing Iterative Closest Point (ICP) for 2D motion estimation. These technical challenges underscore the importance of robust hardware and software integration to address failures in robotics. The experience emphasizes the need for contingency measures, such as alternative SLAM codes, and highlights the value of recording wheel velocity as a backup in case SLAM encounters difficulties.
Page updated
Google Sites
Report abuse