In your last years in engineering, it is natural to develop a capstone project. Waterloo categorizes this into three options:
- Open source
- Research
- Product
The good thing about this is that you are given school time to do this. This means that things are de-risked. Mainly, there is minimal market-risk because it’s not necessary that people need to use this (it’s a school project). This means that you can use all the time you have to technically de-risk something.
Something I’ve noticed about deep-tech companies is that they often spend some time before “starting” by developing the tech out. This is good justification to do a deep-tech startup.
Some ideas:
- Farming autonomomy
- Underwater robots
- Laundry folding robots
- Swarm robot agents
- Truffle detection
- Humanoid robot stuff?
Tasks:
- get back on main branch of lerobot (steven)
-
try best off the shelf model (steven)
- set up simulation (kesu)
- do quest teleop (eddie)
- try non language conditioned model
- dagger on lerobot (krish)
- on trossen