Prototype #3 (2019): 

Prototype #2 (2018): 

Some overview slides: https://docs.google.com/presentation/d/1vobfSEhb0oGUbzrP773AvgAFrEsGztFD6GFTlhZPUSs/edit?usp=sharing

See more details here: https://jgeating.wordpress.com/portfolio/swerve-robot/

Prototype 1: 3 omniwheels with 80mm ginormous [oversized] brushless motors. The wheels were ripped to shreds, and lack of traction control made this very hard to get highly dynamic motions. Was controllable, but nowhere near as agile as a human. 

Prototype 2: 3 wheel swerve drive - Supervised a senior design team to make this one. Same 80mm motors as previous build with 63mm steering motors at a ~4:1 ratio. Friction limited, and iterated on the wheels 3 times before settling on Colson wheels. Still developing, but it was overengineered, super heavy, and unwieldy. Used sliprings with motors mounted directly to the steerable body. 

Prototype 3: four wheel swerve drive - Again, supervised a senior design team to make this one. skateboard hub wheels for the four drive wheels. Focused on making as low profile as possible. Total robot height was ~5in. Used custom sliprings to transmit power and signal to the hub motors. 

VESC motor controllers used for all three prototypes. CAN bus controlled from an Arduino Due. For the last two prototypes, and especially prototype 3, looking at making force sensing footpads for steering while riding on top. First gen used 8 force sensitive resistors. They worked alright, and surprisingly the sensor quality/linearity was okay, but the mechanical robustness just wasn't there. Moved to using 8 load cells instead. Currently integrating this. Also going through an Arduino Due, with 4x Robotshop dual load cell shields to interface to all of them. 

Always interested in collaborating and getting feedback, so hit me up if you're interested. I live in the Boston, MA, USA area. 

9/8/2019 update - metal build of prototype 3 underway: