Visual Inertial Odometry (VIO) is used for estimating pose and trajectory of a system and is a foundational requirement in many emerging applications like AR/VR, autonomous navigation in cars, drones and robots. In this paper, we analyze key compute bottlenecks in VIO and present a highly optimized VIO accelerator based on a hardware-software codesign approach. We detail a set of novel micro-architectural techniques that optimize compute, data movement, bandwidth and dynamic power to make it possible to deliver high quality of VIO at ultra-low latency and power required for budget constrained edge devices. By offloading the computation of the critical linear algebra algorithms from the CPU, the accelerator enables high sample rate IMU usage in VIO processing while acceleration of image processing pipe increases precision, robustness and reduces IMU induced drift in final pose estimate. The proposed accelerator requires a small silicon footprint (1.3 mm2in a 28nm process at 600 MHz), utilizes a modest on-chip shared SRAM (560KB) and achieves 10x speedup over a software-only implementation in terms of image sample-based pose update latency while consuming just 2.2 mW power. In a FPGA implementation, using the EuRoC VIO dataset (VGA 30fps images and 100Hz IMU) the accelerator design achieves pose estimation accuracy (loop closure error) comparable to a software based VIO implementation.

We analyze a class of mechanisms that locomote by switching between constraints. Because of the hybrid nature of such systems, most of the existing analysis tools, developed primarily for smooth systems, can not be directly applied. Our aim is to exploit the special structure provided by Lagrangian mechanics to study the controllability of this class of mechanisms. We base the analysis on a series representation of the evolution of the system. Our main result is a description of trajectories involving switches between constraints at nonzero velocity (impacts) in the presence of large inertial forces (drift). The analysis provides a basis for local motion planning. The results are applied to an example of a two-link planar mechanism that can locomote by clamping one of the links.