Design of an Embedded Multi-Camera Vision System—A Case Study in Mobile Robotics

Valter Costa,Armando Sousa,Peter Cebola,Ana Reis

doi:10.3390/robotics7010012

Abstract

The purpose of this work is to explore the design principles for a Real-Time Robotic Multi Camera Vision System, in a case study involving a real world competition of autonomous driving. Design practices from vision and real-time research areas are applied into a Real-Time Robotic Vision application, thus exemplifying good algorithm design practices, the advantages of employing the “zero copy one pass” methodology and associated trade-offs leading to the selection of a controller platform. The vision tasks under study are: (i) recognition of a “flat” signal; and (ii) track following, requiring 3D reconstruction. This research firstly improves the used algorithms for the mentioned tasks and finally selects the controller hardware. Optimization for the shown algorithms yielded from 1.5 times to 190 times improvements, always with acceptable quality for the target application, with algorithm optimization being more important on lower computing power platforms. Results also include a 3-cm and five-degree accuracy for lane tracking and 100% accuracy for signalling panel recognition, which are better than most results found in the literature for this application. Clear results comparing different PC platforms for the mentioned Robotic Vision tasks are also shown, demonstrating trade-offs between accuracy and computing power, leading to the proper choice of control platform. The presented design principles are portable to other applications, where Real-Time constraints exist.

Highlights

Using cameras as rich sensors, with data coming from vision systems, is a important part of many real world robotic applications, be it by assisting people in their daily tasks [1] or in vision perception related to mobile platforms [2]
Using as motivation the PRO Autonomous Driving competition, the goal of the current research is the search of Real-Time Artificial Vision approaches that are applicable to real robotics problems, such as:
A common application example is autonomous driving and this article presents a scaled down car, in this case, a robotic platform for the autonomous driving competition of the Portuguese Robotics Open

Summary

Introduction

Using cameras as rich sensors, with data coming from vision systems, is a important part of many real world robotic applications, be it by assisting people in their daily tasks [1] or in vision perception related to mobile platforms [2]. One very relevant area of application is autonomous vehicles, representing a major innovation for the automotive industry that, in turn, results in a great economic impact world-wide. In such a dynamic context, work and research in this area has grown greatly over the last years. An autonomous intelligent vehicle has to perform a number of tasks, sometimes in a limited amount of time [6] This involves being able to identify and track road lanes, being able to process traffic lights and road signs, and being consistent at identifying and avoiding obstacles [7,8,9]

Objectives

Methods

Conclusion