Loading system evidence
Loading system evidence
Build a warehouse AMR whose geometry, sensors, odometry, maps, and localisation still hold up once it leaves a clean tutorial.
Sign in to start - freeOutcome
Build
A reliability-tested warehouse AMR with calibration evidence, validated maps, and recovery trials.
Sample incident
Stabilize the robot without freezing joints or reducing the physics rate.
Full syllabus
Connect wheel placement, footprint, collision geometry, and inertial values to observable motion.
Compose map, odom, base, and sensor transforms with correct direction and timing.
Interpret indexes, angles, invalid ranges, scan order, frame IDs, and acquisition time.
Convert ticks and angular velocity into motion estimates with correct signs, units, and timestamps.
Model friction, noise, latency, motor limits, and time so simulation failures transfer to hardware.
Inspect scan matching, odometry priors, loop closures, and map quality as testable evidence.
Interpret particle convergence, covariance, initial pose, sensor models, and relocalisation behavior.
Turn geometry, sensor, odometry, mapping, and localisation assumptions into repeatable tests.
Playgrounds
Small, functional experiments that reveal the math and evidence behind this course’s failures. Open them alongside the labs.
Map LaserScan indexes to physical angles, ranges, invalid values, and rotated sensor frames.
LiDAR angle calculator / LaserScan / ROS 2
Live controlsConvert wheel velocities into robot motion, encoder ticks, and odometry estimates.
differential drive / odometry / encoder ticks
Live controlsFuse wheel odometry and IMU yaw in information form and see how a mis-synced or sign-flipped IMU makes the filter diverge.
EKF / sensor fusion / IMU odometry