Acceleration of contractor algebra on RISCV in the context of mobile robotic
Abstract
Interval arithmetic and contractor algebra are widely used in robotic
as they allow to tackle recurrent tasks like localisation and parameter
estimation in an efficient way. The interval community can rely on
some well known software libraries (ibex, gaol, mpfi ...) which per-
form accurate forward and backward contractions to solve most of the
common problems. However, such libraries always revolve around the
use of low-level maths and fpu functions that are inherently dependent
of the compilation platform and the execution environment. This be-
comes problematic in the area of mobile robotic where the code is often
run on exotic embedded systems which may not support them at all or
produce incorrect behaviours. This area has also its own constraints
when using interval arithmetic. The precision of the computation is
often less of an issue compared to being fast, guaranteed (even when
pessimistic) and energy-efficient. Most interval algorithms based on
contractor algebra expose some data and instruction parallelism. An
hardware implementation specifically tailored for the need of mobile
robotic could be a solution to deal with the aforementioned limitations.
Origin | Files produced by the author(s) |
---|