HyperPPO: A scalable method for finding small policies for robotic control | IEEE Conference Publication | IEEE Xplore