Wayne, wish you would come; you are always a great competitor, with well engineered bots.
I used dead reckoning (wheel encoders & gyros) with accelerometer data to detect collisions (and attempt to get around obstacles). I also had a Lidar-Lite mounted on a servo for scanning the barrels. Lidar worked well in testing and the lidar is rock solid outside and even on the hay bales, but when I got out on the course and tested on Friday/Saturday morning, it appeared that the lidar was freaking out intermittently and spamming the I2C bus, and blocking the gyro signal (a power/noise issue -- my best guess anyway). Didn't always happen, but frequently enough that it was more than likely to mess up a run. So, I unplugged the lidar -- problem solved, but that left getting through the barrels up to a bump and go strategy. With a few less barrels, or if they were a bit further apart, no problem, but they way they were configured, my bots were only able to get through about 1 in 4 tries without getting stuck.
Navigation with the ride-on vehicle was much easier, as there were no obstacles. It was a bit challenging, as I only had a few runs on the course to calibrate (most of the day, the obstacles were on the course -- along with a lot of other vehicles/obstacles).