Embedded Files
Random Arm
Random Arm
Applying random noise to joint rotations and a rating scheme to reach a target. Remembering targets and best paths.
- smoothing: moves 5% towards best results each frame
- fine tuning: each time a solution is found, a joint is locked
- range limits: each joint has rotation range limits (30 degrees)
- randomness: each frame a random set of joint rotations is generated
- rating: distance between end segment and target
- target range: when outside this range, a new target is searched
- best: the best joint rotation is remembered and shown in green
- try: random tries are shown in grey
Manual tracking using history only.
Different twist with simplified code.
Blue has learned thru green's experience.
Blue learns and surpasses real time green by recalling nearest historical best.
Smoothing by adding in 5% of best to real.
Testing winner of rating system: distance between arm end and target.
Best distance visual feedback.
Creating arm segments.
Setting pitch, yaw, roll angles.
Controlling end segments.
Mapping random tries.
3D mapping of random tries.
First segment is attached to home.
Outside range resets best distance, to find new target.
Starting to remember.
Page updated
Report abuse