RESUMO
Autonomous vehicles require efficient self-localisation mechanisms and cameras are the most common sensors due to their low cost and rich input. However, the computational intensity of visual localisation varies depending on the environment and requires real-time processing and energy-efficient decision-making. FPGAs provide a solution for prototyping and estimating such energy savings. We propose a distributed solution for implementing a large bio-inspired visual localisation model. The workflow includes (1) an image processing IP that provides pixel information for each visual landmark detected in each captured image, (2) an implementation of N-LOC, a bio-inspired neural architecture, on an FPGA board and (3) a distributed version of N-LOC with evaluation on a single FPGA and a design for use on a multi-FPGA platform. Comparisons with a pure software solution demonstrate that our hardware-based IP implementation yields up to 9× lower latency and 7× higher throughput (frames/second) while maintaining energy efficiency. Our system has a power footprint as low as 2.741 W for the whole system, which is up to 5.5-6× less than what Nvidia Jetson TX2 consumes on average. Our proposed solution offers a promising approach for implementing energy-efficient visual localisation models on FPGA platforms.
RESUMO
Autonomous vehicles require precise and reliable self-localization to cope with dynamic environments. The field of visual place recognition (VPR) aims to solve this challenge by relying on the visual modality to recognize a place despite changes in the appearance of the perceived visual scene. In this paper, we propose to tackle the VPR problem following a neuro-cybernetic approach. To this end, the Log-Polar Max-Pi (LPMP) model is introduced. This bio-inspired neural network allows building a neural representation of the environment via an unsupervised one-shot learning. Inspired by the spatial cognition of mammals, visual information in the LPMP model are processed through two distinct pathways: a "what" pathway that extracts and learns the local visual signatures (landmarks) of a visual scene and a "where" pathway that computes their azimuth. These two pieces of information are then merged to build a visuospatial code that is characteristic of the place where the visual scene was perceived. Three main contributions are presented in this article: 1) the LPMP model is studied and compared with NetVLAD and CoHog, two state-of-the-art VPR models; 2) a test benchmark for the evaluation of VPR models according to the type of environment traveled is proposed based on the Oxford car dataset; and 3) the impact of the use of a novel detector leading to an uneven paving of an environment is evaluated in terms of the localization performance and compared to a regular paving. Our experiments show that the LPMP model can achieve comparable or better localization performance than NetVLAD and CoHog.
