Page 430 - Automotive Engineering Powertrain Chassis System and Vehicle Body

14.2
Chapter 14.2

Decisional architecture

Ljubo Vlacic and M. Parent

14.2.1 Introduction a structured environment with motion rules (the high-
way code) and it is possible to take advantage of these
Autonomy in general and motion autonomy in particular features in order to design a control architecture that is
has been a long standing issue in robotics. In the late 1960s efficient, robust and flexible.
and early, 1970s, Shakey (Nilsson, 1984) was one of the This chapter is organized as follows: in Section 14.2.2,
first robots able to move and perform simple tasks an overview of the existing approaches to implementing
autonomously. Ever since, many authors have proposed a control architecture is presented and discussed; this
control architectures to endow robot systems with various section describes the three main classes of existing
autonomous capabilities. Some of these architectures are approaches (deliberative, reactive and hybrid), along
reviewed in Section 14.2.2 and compared to the one with their main implementation alternatives and their
presented in Section 14.2.3. These approaches differ in respective advantages and drawbacks.
several ways, however it is clear that the control structure Section 14.2.3 describes the hybrid control architec-
of an autonomous robot placed in a dynamic and partially ture developed by Sharp. The rationale of the architec-
known environment must have both deliberative and re- ture and its main features are overviewed in Section
active capabilities. In other words, the robot should be 14.2.3.1. This section introduces the key concept of the
able to decide which actions to carry out according to its sensor-based manoeuvre, i.e. general templates that
goal and current situation; it should also be able to take encode the knowledge of how a specific motion task is to
into account events (expected or not) in a timely manner. be performed. The models of the experimental car-like
The control architecture presented in this chapter vehicles that are used throughout the chapter are then
aims at meeting these two requirements. It is designed to described in Section 14.2.3.2. Afterwards the concept
endow a car-like vehicle moving on the road network of the sensor-based manoeuvre is explored in Section
with motion autonomy. It was initially developed within 14.2.3.3 and three types of manoeuvres are presented in
the framework of the French Praxite `le programme aimed detail (Sections 14.2.3.4, 14.2.3.5 and 14.2.3.6). These
at the development of a new urban transportation system manoeuvres have been implemented and successfully
based on a fleet of electric vehicles with autonomous tested on our experimental vehicles; the results of these
motion capabilities (Daviet and Parent, 1996); more experiments are finally presented in the Section 14.2.4.
recent work on this topic is done within the framework One important component of the architecture is the
of the LaRA (Automated Road) French project. The motion planner whose purpose is to determine the tra-
road network is a complex environment; it is partially jectory leading the vehicle to its goal. Motion planning for
known and highly dynamic with moving obstacles (other car-like vehicles in dynamic environments remains an
vehicles, pedestrians, etc.) whose future behaviour is open problem and a practical solution to this intricate
not known in advance. However the road network is problem is presented in Section 14.2.5.

Intelligent Vehicle Technologies; ISBN: 9780750650939
Copyright Ó 2001 Ljubo Vlacic and M. Parent. All rights of reproduction, in any form, reserved.