피지컬 AI Lab https://validator.w3.org/feed/docs/rss2.html robosuite composite actions: check each robot part’s slice MuJoCo rangefinder feedback: from sensor data to control MuJoCo actuator_force vs qfrc_actuator: why joint torque differs MuJoCo contacts: margin, gap and explicit pairs OSC delta targets: achieved position or previous goal? RMF graph-merge tolerances: matching a pose to nearby lanes Joining robot paths: a shared endpoint is only the first check 피지컬 AI Lab Transforming a wrench between sensor and tool origins Overlapping bounding boxes do not prove a mesh collision XYZ Euler gimbal lock: why roll and yaw stop being independent Equal Bezier parameter steps do not mean equal travel Four supports do not guarantee equal load sharing LiDAR scan spacing: a narrow post can fall between sample directions Low-speed encoder velocity: zero counts does not mean a stopped wheel IMU lever arms: matching orientation does not mean matching acceleration LiDAR scan deskew: one pose can bend a straight wall Scaled mesh normals: normalization cannot fix the wrong transform Overhung shaft loads: why one support reaction exceeds the load Gearbox backlash: the same input angle can leave two output positions Short servo moves: raising the speed limit may change nothing Stereo depth on repeating stripes: a perfect match can be the wrong match Why two gripper point contacts cannot resist every torque Blender ray casting: a hit is not an inside test Why a blind-hole screw can bottom before its head seats Geneva drives: a stopped animation does not prove a locked dwell Moving pulleys: angled cable legs change the travel ratio Wheel Odometry Slip and Covariance Robot Coordinate Frames and TF2: map, odom, base and tool Three-marker coordinate frames: why nearly aligned points make the axes swing Quaternion interpolation: the same endpoint can take a different path Why a rotated Blender child skews under a scaled parent Camera depth vs range: the ray-normalization mistake that bends a plane Hand-eye calibration: why pure translations cannot determine camera offset Ackermann steering: why equal front-wheel angles miss the turn center V-blocks: changing shaft diameter changes center height Differential-drive odometry: why averaging wheel travel alone misses the turn Mirrored robot meshes: why a quaternion cannot carry the reflection Planar arm Jacobians: check tip derivatives before mapping joint rates Rigid-body inertia: check the center and axes before copying Blender camera clipping: why a nearby mesh can disappear Stereo depth: why one pixel matters more at long range Robot TCP offsets: keeping the tool tip in place Chain sprockets: constant rpm, varying chain speed Multi-start lead screws: pitch versus travel per revolution Planetary gears: planet spin versus carrier rotation Peg insertion: how bore depth changes clearance under tilt Cable-carrier U-loops: the end-wall clearance a centerline check misses Two universal joints: check yoke phasing and bend-angle equality separately 글 목록