My custom-built mobile robot is running ROS2 humble; tele-op through a joystick is rock solid, and the hardware stack (motors, encoders, power, safety relays) has already proved itself in long bench tests. Mapping, however, is still an issue. With the current LIDAR-only SLAM setup the occupancy grid comes out distorted, misaligned and visibly warped, so the global planner has nothing reliable to work with. Wheel odometry has been calibrated thoroughly, so I’m leaning toward a problem in the TF chain, bad sensor fusion defaults or simply mis-tuned parameters in our chosen SLAM package (slam_toolbox, though I’m open to cartographer or another option if you see the benefit). I’d like you to step in, trace the root cause, and deliver a clean map that matches the physical environment closely enough for repeatable autonomous navigation. What you’ll hand over: • Corrected SLAM/TF configuration files and any launch changes • A short report summarising the changes and the reasoning behind them • Evidence the fix works: either a rosbag + resulting map files or a brief screen-capture demo showing an accurate live map while the robot is driving Remote access to the robot’s workstation (SSH + screen sharing) is available, along with recent rosbags, URDF, and the full build tree. If you need extra diagnostics or specific test maneuvers I can drive and record them for you. Let me know your preferred workflow and how soon you can begin; I’m ready to give rapid feedback and test any iterative tweaks you propose.