Where the method was learned.
Graduate and undergraduate work, 2014–2021. Nearly all of it is the same shape: build a model of something that will not behave on its own, then close a loop around it until it does. The hardware changes; the method does not. My professional work is on the experience page.
Inverted pendulum
Michigan, 2019–21Swing-up and stabilisation of a multi-degree-of-freedom inverted pendulum on hardware I built the driver for: bring the rod up from hanging, then hold it upright against whatever you do to it.
Read more →Magnetic levitation
Michigan, 2019–21Holding a steel nut suspended in mid-air in an electromagnetic field — a plant with no stable resting point in either direction.
Read more →Adaptive cruise control
Michigan, 2019–21A low-level adaptive cruise controller with automatic steering and haptic feedback, running on an NXP S32K with sensor and actuator traffic carried over CAN.
Read more →Predictive control in mixed traffic
Michigan, 2019–21A predictive controller that lets an autonomous vehicle use traffic data broadcast by the cars around it — 26% less energy and 5% less travel time over a 10 km route.
Read more →Self-driving toolkit
Personal, 2020Ten projects working up from OpenCV lane detection to YOLOv3 object detection and 43-class traffic-sign recognition — the perception half of everything else on this page.
Read more →SAE BAJA all-terrain vehicle
Undergraduate, 2016–18Three years on a 25-person team building a CVT-driven all-terrain buggy for SAE BAJA — team member, then Design and CAE Head, then SAE Collegiate President. My part was the chassis.
Read more →Steering control
Michigan, 2019–21A multi-order controller for vehicle steering, designed against a model of the steering system rather than tuned on the car.
Read more →