23 October 2023 Weekly Meeting

[Woolfrey]

Chair:

• Jon @Woolfrey

Attendees:

• Stefano @steb6
• Andrea @andrearosasco
• Vignesh @vigisushrutha23

Absent:

• Simone @SimoneMic (on vacation)

---

Agenda:

Note: Last week's items should become this week's agenda.

Item	Tag	Responsible	Comment
Test more parameters for singularity avoidance	#35	@Woolfrey	Busy working on model predictive control code / trajectory optimisation. I suspect the control frequency can improve control barrier function.
Work on trajectory generator for MPC	N/A	@Woolfrey	Solved trajectory optimisation. Typed up notes (attached in PDF below).
Plan for final demonstration	N/A	@Woolfrey	Busy last week with trajectory optimisation; Simone away this week, better to do it next week. (Maybe Tuesday?)
Implement experiments for integration paper	N/A	@vigisushrutha23	Did experiments with Simone walking past the robot. Penalty on joint motion in AMI's walking controller overrides bimanual control, causing collision.

Additional Discussion Points:

Anything not captured in the agenda.

Discussion Point	Raised By	Tag	Outcome
Need to organise time to plan for end-of-year demo.	@Woolfrey	N/A	Earliest Tuesday next week, as Simone is on vacation this week.

---

Next Meeting:

Note: These will become the agenda for the next meeting.

Item	Tag	Responsible
Finished behaviour tree for manipulation from navigation	N/A	@vigisushrutha23
Build docker for navigation & manipulation	N/A	@vigisushrutha23
Implement mobile robot trajectory in C++	N/A	@Woolfrey
Test vision and BT in Gazebo with newest manipulation code and push to demo branch	N/A	@andrearosasco @steb6 @Woolfrey
Develop pre-demo checklist	N/A	@Woolfrey @vigisushrutha23 @steb6 @andrearosasco

[Woolfrey]

Trajectory optimisation for wheeled mobile robot using the calculus of variations:

Trajectory_Optimisation_for_Mobile_Robots.pdf

[Woolfrey]

Root Cause Analysis

5 Whys

Repeatedly ask "why?" about problems until a fundamental issue can be identified.

Outcomes identified:

• Develop a pre-demo checklist to quickly identify problems, and revert back to old code that we know is working (need to keep track of which commit to use for each module)
• Push local code to a "last-working-demo" branch after every demonstration.