LouKordos walking_controller issues

LouKordos / walking_controller

The main walking controller code for the Bipedal Robot.

13 stars 2 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Add Github Actions example

#92 LouKordos closed 3 years ago
0
Add CI using Github Actions

#91 LouKordos closed 3 years ago
0
Use custom base image containing deps for faster auto builds

#90 LouKordos closed 3 years ago
0
Reliable Walking in Gazebo Bullet, Web UI for real-time control

#89 LouKordos closed 3 years ago
0
Implement Web UI connection for real-time robot control (#33)

#88 LouKordos closed 3 years ago
0
Transform state and control action weights from body frame to world frame (see #85)

#87 LouKordos closed 3 years ago
0
Add support for running on Windows using WSLg

#86 LouKordos opened 3 years ago
1
Q (state gains) and R (control action gains) are in world frame, causing them to be inaccurate for nonzero psi

#85 LouKordos closed 3 years ago
4
Make logging asynchronous to fix #80

#84 LouKordos closed 3 years ago
0
Implement graceful program exit by catching CTRL + C

#83 LouKordos closed 3 years ago
0
Move contact and trajectory planner updates to Leg thread running at 1kHz, log execution time for each code block, add wrapper functions for threadsafe variable access

#82 LouKordos closed 3 years ago
0
Foot sometimes hovers above ground for short durations because stance phase takes over too late, causing the foot to suddenly slam into the ground

#81 LouKordos opened 3 years ago
0
Code block takes unusally long time at random points in simulation and causes the robot to break

#80 LouKordos closed 3 years ago
3
Send Center of Mass State to Leg Threads at 1kHz instead of using 50Hz MPC state (Fix #77)

#79 LouKordos closed 3 years ago
0
Replace iteration-based contact planner on MPC thread with stateless time-based contact planner (Fix #73)

#78 LouKordos closed 3 years ago
0
Leg control threads use state received by MPC thread, which is updated much less frequently

#77 LouKordos closed 3 years ago
1
Trajectory Planner only uses single latest state for transforming desired position into body frame

#76 LouKordos opened 3 years ago
1
Time-based Contact Planner timing is not working as expected

#75 LouKordos closed 3 years ago
0
Somehow also save / copy state.log to log directory

#74 LouKordos closed 3 years ago
1
Gait/Contact Planner runs on MPC thread and is thus delayed by solver time delays

#73 LouKordos closed 3 years ago
1
Fix compile caching in Docker container to reduce build time

#72 LouKordos opened 3 years ago
1
Constraint for limiting step length based on velocity error breaks with negative desired Y velocity

#71 LouKordos closed 3 years ago
1
Desired foot position formula does not account for steady state position error

#70 LouKordos closed 3 years ago
1
Wrapper script for Docker run and compile and backing up snapshots of source code

#69 LouKordos opened 3 years ago
4
Seemingly regular spikes in solver time

#68 LouKordos closed 3 years ago
2
Autogenerate C++ Header from Jupyter Notebook parameters along with NLP binary

#67 LouKordos opened 3 years ago
0
Improve docs / README

#66 LouKordos opened 3 years ago
0
Handle singularities in joint space to avoid flipping leg joints

#65 LouKordos opened 3 years ago
0
Try JIT compilation for NLP, maybe this results in faster evaluations

#64 LouKordos opened 3 years ago
0
Try plotly for more interactive plots in Python Code

#63 LouKordos opened 3 years ago
0
Use quaternions where possible to avoid Gimbal Lock

#62 LouKordos opened 3 years ago
0
Log IPOPT solver info to file

#61 LouKordos opened 3 years ago
1
Slip detection by keeping track of deviation from touchdown position

#60 LouKordos opened 3 years ago
0
Implement delay compensation in a better way to avoid accidentally using x_t instead of the compensated state

#59 LouKordos opened 3 years ago
0
Initialize the solver better somehow to fix first iteration time

#58 LouKordos closed 3 years ago
1
Test integrating the leg model in C++ to gain insight about future states

#57 LouKordos opened 3 years ago
0
Move nanosleep from plugin to controller to be able to log actual iteration times

#56 LouKordos opened 3 years ago
0
Foot placement seemingly suboptimal

#55 LouKordos closed 3 years ago
1
Check Joint orientation and possible sign swaps in simulation

#54 LouKordos closed 3 years ago
1
Use configuration file for tuning parameters etc.

#53 LouKordos opened 3 years ago
0
Find a way of linking gains used in nlp.so to file metadata

#52 LouKordos closed 3 years ago
0
Consider constraining force based on leg configuration and needed torques instead of constant forces

#51 LouKordos opened 3 years ago
1
Consider 4-point contact model for foot

#50 LouKordos opened 3 years ago
0
Kalman Filter for State Estimation + Noise Testing in Gazebo

#49 LouKordos opened 3 years ago
0
Use RViz to visualize controller data

#48 LouKordos opened 3 years ago
2
Consider running average dt for MPC combined with adaptive prediction horizon length

#47 LouKordos opened 3 years ago
0
Include foot positioning into optimization problem to give the solver more control authority

#46 LouKordos opened 3 years ago
1
Consider seperating state update from rest of code

#45 LouKordos opened 3 years ago
0
Use Vector for `r` instead of seperate component values

#44 LouKordos opened 3 years ago
0
Test suite with gzserver docker container to validate a build

#43 LouKordos opened 3 years ago
5