Dynamic parameter identification code for rokae xmate manipulator based on MATLAB, including excitation trajectory optimization, LSM method, and Recursive Newton-Euler Algorithm.
- Formulate robot dynamic model through Newton-Euler method.
- Linearize robot dynamic model to obtain regressor.
- Apply QR decomposition to obtain minimal regressor.
- Convert minimal parameter set Pmin to standard set.
- Verify error between observation and NE-based estimation.
- Deduce robot dynamic equation item.
See ./dynamics/README.md
and run_dynamics.m
for details.
- optimize excitation trajectory based on cond of minimal regressor.
- matrixized constraints for trajectory optimization: |q| < qmax, |qd| < qdmax, |qdd| < qddmax, q0=qn=q_init
- plot figures of {q, qd, qdd, cart_pos} and also animation.
- cpp scripts for running excitation trajectory based on rci client.
See ./excitation/README.md
and run_optimize.m
for details.
- Downsample observation data to assigned size.
- Apply butterworth and zero-phase filter to q, qd, qdd, tau data.
- Plot figures of raw and filtered data.
See ./filtering/README.md
and run_filtering.m
for details.
- Apply LSE (least square estimation) to figure out minimal param set.
- Verify error between observation and estimation by min regressor.
See ./identify/README.md
and run_identify.m
for details.
- Derive robot dynamics, regressor and minimum paramset:
run_dynamics.m
PART-IA and PART-IB. - Optimize excitation trajectory:
run_optimze.m
. - Data filtering and processing:
run_filtering.m
. - Estimate minimum paramset using LSE:
run_identify.m
. - Map minimum paramset to standard paramset:
run_dynamics.m
PART-II. - [OPTIONAL] Test the performance of identified dynamics model with standard paramset:
run_dynamics.m
PART-III.
Notes:
- Keep System of Units consistent throughout the project (mm and Nmm).
- Additional adjustment of virtual mass in
dyn_mapping_Pmin2P.m
is needed for better paramset mapping.
Obtain min regressor matrix in \dynamics
and then turn to \excitation
.
Copy raw sensor data in \filtering
and then turn to \dynamics
.
[1] Yanjun Liu. "Study on Parameter Identification for Rokae XB4," 2019. link.
[2] Craig, John J. "Introduction to Robotics," 2005.