Physical AI Lab

Plan robot grasps from uncertain object geometry through contact scoring, IK, collision filtering, approach, lift, tactile checks and held-out pick tests.

Evaluate robot VLA policies with reproducible task strata, progress metrics, confidence intervals, latency, shift tests, interventions and failure analysis.

Understand robot action chunking, ACT, temporal ensembling, chunk and execution horizons, imitation data, latency, disturbances and safety tests.

Learn robot diffusion policy action denoising, observation and action horizons, receding-horizon execution, latency, training data and safety tests.

Identify robot dynamics with safe excitation, synchronized data, estimable parameters, physical constraints, residual analysis and held-out validation.

Compare robot collision detection using torque sensors, motor current and disturbance observers, thresholds, detection time and stop validation.

Design ROS 2 real-time robot control with bounded deadlines, scheduling, memory, ros2_control interfaces, tracing and worst-case jitter tests.

Tune robot joint PID and cascaded current, velocity and position loops using response, saturation, noise, resonance, payload and thermal evidence.

Compare EtherCAT and CAN FD robot networks by control cycle, clock synchronization, bus load, topology, diagnostics, wiring and fault behavior.

Compare robot LiDAR, visual and visual-inertial SLAM by sensors, loop closure, calibration, map output, metrics and failure conditions.