In preparation for Istanbul, we’ve begun overhauling the sensor filtering pipeline of our motion engine. Currently (see the graph) things are pretty insane, with three layers of filtering before any data makes it into the control algorithms. With any luck, we’ll be able to tune the early stages of filtering to reduce any unnecessary delay.
Jack, Nathan, Ben, Brian and Chown-dog traveled down to Washington DC this weekend for the USA Science and Engineering Festival. We’re on the national mall with a great view of the
White House capitol and the Washington Monument. More pictures and details to come later!
The Northern Bites represented in Singapore, making the quarter-final round despite numerous technical difficulties (our router was electrocuted within a half hour of arriving on scene, don’t even mention wireless to Chown-dog…).
Back on campus, I (Nathan) have been busy working on an optimizer system for our gaits. Eventually we’ll be able to place the robot on a foreign carpet and, using a combination of accelerometers, foot sensor data and a hacked together ball distance model the robot will self-select a stable, fast gait. En route to this goal there will be slightly more heavy-lifting inside the motion engine than I had originally planned. Aldebaran has updated their documention on the Nao’s weights and offsets and our code badly needs an update to reflect the new robot version.
To keep you all interested, here’s some video (simulator & real-life) of one of the first stable gaits the optimizer has produced for us thus far. Yes, the Nao is barely shuffling along, but remember that 100% of the gait was machine-generated. We’ll get there.