On the Tradeoff Between Hardware Protection and Optimization Success: A Case Study in Onboard Evolutionary Robotics for Autonomous Parallel Parking (original) (raw)

The Effect of Fitness Function Design on Performance in Evolutionary Robotics

Proceedings of the 2015 on Genetic and Evolutionary Computation Conference - GECCO '15, 2015

Fitness function design is known to be a critical feature of the evolutionaryrobotics approach. Potentially, the complexity of evolving a successful controller for a given task can be reduced by integrating a priori knowledge into the fitness function which complicates the comparability of studies in evolutionary robotics. Still, there are only few publications that study the actual effects of different fitness functions on the robot's performance. In this paper, we follow the fitness function classification of and investigate a selection of four classes of fitness functions that require different degrees of a priori knowledge. The robot controllers are evolved in simulation using NEAT and we investigate different tasks including obstacle avoidance and (periodic) goal homing. The best evolved controllers were then post-evaluated by examining their potential for adaptation, determining their convergence rates, and using cross-comparisons based on the different fitness function classes. The results confirm that the integration of more a priori knowledge can simplify a task and show that more attention should be paid to fitness function classes when comparing different studies.

Racing to improve on-line, on-board evolutionary robotics

Genetic and Evolutionary Computation Conference, GECCO'11, 2011

In evolutionary robotics, robot controllers are often evolved in a separate development phase preceding actual deployment -we call this off-line evolution. In on-line evolutionary robotics, by contrast, robot controllers adapt through evolution while the robots perform their proper tasks, not in a separate preliminary phase. In this case, individual robots can contain their own self-sufficient evolutionary algorithm (the encapsulated approach) where individuals are typically evaluated by means of a time sharing scheme: an individual is given the run of the robot for some amount of time and fitness corresponds to the robot's task performance in that period.

Fitness functions in evolutionary robotics: A survey and analysis

Robotics and Autonomous Systems, 2009

This paper surveys fitness functions used in the field of evolutionary robotics (ER). Evolutionary robotics is a field of research that applies artificial evolution to generate control systems for autonomous robots. During evolution, robots attempt to perform a given task in a given environment. The controllers in the better performing robots are selected, altered and propagated to perform the task again in an iterative process that mimics some aspects of natural evolution. A key component of this process -one might argue, the key component -is the measurement of fitness in the evolving controllers. ER is one of a host of machine learning methods that rely on interaction with, and feedback from, a complex dynamic environment to drive synthesis of controllers for autonomous agents. These methods have the potential to lead to the development of robots that can adapt to uncharacterized environments and which may be able to perform tasks that human designers do not completely understand. In order to achieve this, issues regarding fitness evaluation must be addressed. In this paper we survey current ER research and focus on work that involved real robots. The surveyed research is organized according to the degree of a priori knowledge used to formulate the various fitness functions employed during evolution. The underlying motivation for this is to identify methods that allow the development of the greatest degree of novel control, while requiring the minimum amount of a priori task knowledge from the designer.

Embodied, on-line, on-board evolution for autonomous robotics

2010

Artificial evolution plays an important role in several robotics projects. Most commonly, an evolutionary algorithm (EA) is used as a heuristic optimiser to solve some engineering problem, for instance an EA is used to find good robot controller. In these applications the human designers/experimenters orchestrate and manage the whole evolutionary problem solving process and incorporate the end result -that is, the (near-)optimal solution evolved by the EA-into the system as part of the deployment. During the operational period of the system the EA does not play any further role. In other words, the use of evolution is restricted to the pre-deployment stage.

Self-adapting fitness evaluation times for on-line evolution of simulated robots

Proceeding of the fifteenth annual conference on Genetic and evolutionary computation conference - GECCO '13, 2013

This paper is concerned with on-line evolutionary robotics, where robot controllers are being evolved during a robots' operative time. This approach offers the ability to cope with environmental changes without human intervention, but to be effective it needs an automatic parameter control mechanism to adjust the evolutionary algorithm (EA) appropriately. In particular, mutation step sizes (σ) and the time spent on fitness evaluation (τ ) have a strong influence on the performance of an EA. In this paper, we introduce and experimentally validate a novel method for self-adapting τ during runtime. The results show that this mechanism is viable: the EA using this self-adaptative control scheme consistently shows decent performance without a priori tuning or human intervention during a run.

Evolutionary Robotics: Exploring New Horizons

New Horizons in …, 2011

This paper considers the field of Evolutionary Robotics (ER) from the perspective of its potential users: roboticists. The core hypothesis motivating this field of research is discussed, as well as the potential use of ER in a robot design process. Four main aspects of ER are presented: (a) ER as an automatic parameter tuning procedure, which is the most mature application and is used to solve real robotics problem, (b) evolutionary-aided design, which may benefit the designer as an efficient tool to build robotic systems (c) ER for online adaptation, i.e. continuous adaptation to changing environment or robot features and (d) automatic synthesis, which corresponds to the automatic design of a mechatronic device and its control system. Critical issues are also presented as well as current trends and pespectives in ER. A section is devoted to a roboticist's point of view and the last section discusses the current status of the field and makes some suggestions to increase its maturity. S. Doncieux et al. (Eds.): New Horizons in Evolutionary Robotics, SCI 341, pp. 3-25.

Evolutionary robotics: A survey of applications and problems

Evolutionary Robotics, 1998

This paper reviews evolutionary approaches to the automatic design of real robots exhibiting a given behavior in a given environment. Such a methodology has been successfully applied to various wheeled and legged robots, and to numerous behaviors including wall-following, obstacle-avoidance, light-seeking, arena cleaning and target seeking. Its potentialities and limitations are discussed in the text and directions for future work are outlined.

Controlling maximum evaluation duration in on-line and on-board evolutionary robotics

Evolving Systems, 2014

On-line evolution of robot controllers allows robots to adapt while they perform their proper tasks. In our investigations, robots contain their own self-sufficient evolutionary algorithm (known as the encapsulated approach) where individual solutions are evaluated by means of a time sharing scheme: an individual controller is given the run of the robot for some amount of time and fitness corresponds to the robot's task performance during that period. In this paper, we propose and provide a detailed analysis of two on-the-fly control schemes to set the evaluation time in highly dynamic scenarios with completely different tasks. One scheme, called the roulettewheel selection scheme, stochastically selects evaluation time from promising intervals similar to multi-armed bandit schemes. The other scheme, named Heuristic-Rule (H-Rule), tweaks the evaluation time using specific heuristics. Our experiments show that H-Rule gives stable performance in different scenarios and can serve as a viable alternative to pre-selected optimal evaluation time.

The Transferability Approach: Crossing the Reality Gap in Evolutionary Robotics

2011

Abstract The reality gap, that often makes controllers evolved in simulation inefficient once transferred onto the physical robot, remains a critical issue in Evolutionary Robotics (ER). We hypothesize that this gap highlights a conflict between the efficiency of the solutions in simulation and their transferability from simulation to reality: the most efficient solutions in simulation often exploit badly modeled phenomena to achieve high fitness values with unrealistic behaviors.

Noise and the reality gap: The use of simulation in evolutionary robotics

Lecture Notes in Computer Science, 1995

The pitfalls of naive robot simulations have been recognised for areas such as evolutionary robotics. It has been suggested that carefully validated simulations with a proper treatment of noise may overcome these problems. This paper reports the results of experiments intended to test some of these claims. A simulation was constructed of a two-wheeled Khepera robot with IR and ambient light sensors. This included detailed mathematical models of the robot-environment interaction dynamics with empirically determined parameters. Arti cial evolution was used to develop recurrent dynamical network controllers for the simulated robot, for obstacle-avoidance and light-seeking tasks, using di erent levels of noise in the simulation. The evolved controllers were down-loaded onto the real robot and the correspondence between behaviour in simulation and in reality was tested. The level of correspondence varied according to how much noise was used in the simulation, with very good results achieved when realistic quantities were applied. It has been demonstrated that it is possible to develop successful robot controllers in simulation that generate almost identical behaviours in reality, at least for a particular class of robot-environment interaction dynamics.