Andrew Price is a Canadian born machine learning researcher and
Aerospace engineer. He has lived, studied and worked in Canada, Japan,
Sri Lanka and Switzerland. He is a native English speaker but can get
by with French or Japanese in informal settings. He is PADI open water
diver certified. He is married and has two annoying boys.
His research interests include machine learning, robotics, signal
processing and orbital dynamics. He has experience with spacecraft
qualification testing, in-flight measurements, environmental noise
controllers, dataset image rendering, machine learning and network
compression. Links to the various relevant projects can be found on
this site under "Projects".
Module-Wise network Quantization for 6D Object Pose Estimation.
This project employed machine learning to perform Six Degree of
Freedom (6DoF) pose estimation on a number of different
datasets. We then compressed the network using our novel
module-wise compression approach. In certain cases, we were
able to demonstrate higher accuracy with a network compressed
by a factor of 32X. This is highly promising for 6DoF pose
estimation edge computing on spacecraft.