What I build

• Scout 2.0 + xArm6 fruit recycling robot: YOLO RGB-D detection for [N] classes (orange, carrot, apple, ball), workspace filtering, MoveIt pick and place, ArUco bin mapping (IDs 0–3), FSM with grasp retry up to [R] times and [S] scan steps

• Pose-based EKF SLAM: encoder prediction with ICP relative pose updates, Mahalanobis gating with threshold [T], pose cloning and pruning up to [K] poses

• Task-priority kinematic control: recursive null-space control with inequality tasks (joint limits, obstacle avoidance) and DLS damping [λ]

• Next-best-view exploration: sampling-based NBV scoring with Dubins turning radius [ρ] and path resolution [Δs]

• Longitudinal dermoscopy (iToBoS): segmentation + change assessment with ΔE color metrics (CIEDE2000) and temporal growth modeling (PINN, reaction–diffusion)