Gianpiero Francesca

AutoMoDe: A novel approach to the automatic design of control software for robot swarms

We introduce AutoMoDe: a novel approach to the automatic design of control software for robot swarms. The core idea in AutoMoDe recalls the approach commonly adopted in machine learning for dealing with the bias–variance tradedoff: to obtain suitably general solutions with low variance, an appropriate design bias is injected. AutoMoDe produces robot control software by selecting, instantiating, and combining preexisting parametric modules—the injected bias. The resulting control software is a probabilistic finite state machine in which the topology, the transition rules and the values of the parameters are obtained automatically via an optimization process that maximizes a task-specific objective function. As a proof of concept, we define AutoMoDe-Vanilla, which is a specialization of AutoMoDe for the e-puck robot. We use AutoMoDe-Vanilla to design the robot control software for two different tasks: aggregation and foraging. The results show that the control software produced by AutoMoDe-Vanilla (i) yields good results, (ii) appears to be robust to the so called reality gap, and (iii) is naturally human-readable.

Ant Colony Optimization for Deadlock Detection in Concurrent Systems

Ensuring deadlock freedom is one of the most critical requirements in the design and validation of concurrent systems. The biggest challenge toward the development of effective deadlock detection schemes remains the state-space explosion problem when model checking is used for proving the correctness of a system with respect to a desired behavior. In this paper we propose the use of the Ant Colony Optimization (ACO) to reduce the state explosion problem arising when finding deadlocks in complex networks described using Calculus of Communicating Systems (CCS). Moreover, ACO is used to provide minimal counterexamples. In fact, although one of the strongest advantages of model checking is the generation of counterexamples when verification fails, traditional model checkers may return very long counterexamples. We present an implementation of our technique and encouraging experimental results on several benchmarks. These results are then compared with other heuristic-based search strategies, retaining the advantages of our approach.

Automatic Design of Robot Swarms: Achievements and Challenges

Automatic design is a promising approach to the design of control software for robot swarms. In an automatic design method, the design problem is cast into an optimization problem and is addressed using an optimization algorithm. In this article, we review studies in which automatic design methods are successfully applied. In particular, we focus our attention on how automatic methods are empirically assessed. An apparent issue that emerges from our review is that a solid, well- established, and consistently applied empirical practice is still missing. For example, studies that propose new methods and ideas do not typically provide any comparison with existing ones. We maintain that the lack of a proper empirical practice hinders the progress of the domain. In this article, we pursue two goals: we highlight the notable achievements in the automatic design of control software for robot swarms and we discuss the challenges to be overcome to establish a proper empirical practice for the domain.

Analysing an Evolved Robotic Behaviour Using a Biological Model of Collegial Decision Making

Evolutionary robotics can be a powerful tool in studies on the evolutionary origins of self-organising behaviours in biological systems. However, these studies are viable only when the behaviour of the evolved artificial system closely corresponds to the one observed in biology, as described by available models. In this paper, we compare the behaviour evolved in a robotic system with the collegial decision making displayed by cockroaches in selecting a resting shelter. We show that artificial evolution can synthesise a simple self-organising behaviour for a swarm of robots, which presents dynamics that are comparable with the cockroaches behaviour.

The TAM: abstracting complex tasks in swarm robotics research

Research in swarm robotics focuses mostly on how robots interact and cooperate to perform tasks, rather than on the details of task execution. As a consequence, researchers often consider abstract tasks in their experimental work. For example, foraging is often studied without physically handling objects: the retrieval of an object from a source to a destination is abstracted into a trip between the two locations—no object is physically transported. Despite being commonly used, so far task abstraction has only been implemented in an ad hoc fashion. In this paper, we propose a new approach to abstracting complex tasks in swarm robotics research. This approach is based on a physical device called the “task abstraction module” (TAM) that abstracts single-robot tasks to be performed by an e-puck robot. A complex multi-robot task can be abstracted using a group of TAMs by first modeling the task as the set of its constituent single-robot subtasks and then abstracting each subtask with a TAM. We present a collection of tools for modeling complex tasks, and a framework for controlling a group of TAMs such that the behavior of the group implements the model of the task. The TAM enables research on cooperative behaviors and complex tasks with simple, cost-effective robots such as the e-puck—research that would be difficult and costly to conduct using specialized robots or ad hoc task abstraction. We demonstrate how to abstract a complex task with multiple TAMs in an example scenario involving a swarm of e-puck robots.

An Experiment in Automatic Design of Robot Swarms

We present an experiment in automatic design of robot swarms. For the first time in the swarm robotics literature, we perform an objective comparison of multiple design methods: we compare swarms designed by two automatic methods—vanilla and EvoStick—with swarms manually designed by human experts. vanilla and EvoStick have been previously published and tested on two tasks. To evaluate their generality, in this paper we test them without any modification on five new tasks. Besides confirming that vanilla is effective, our results provide new insight into the design of robot swarms. In particular, our results indicate that, at least under the adopted experimental protocol, not only does automatic design suffer from the reality gap, but also manual design. The results also show that both manual and automatic methods benefit from bias injection. In this work, bias injection consists in restricting the design search space to the combinations of pre-existing modules. The results indicate that bias injection helps to overcome the reality gap, yielding better performing robot swarms.

Off-line and on-line tuning: a study on operator selection for a memetic algorithm applied to the QAP

Tuning methods for selecting appropriate parameter configurations of optimization algorithms have been the object of several recent studies. The selection of the appropriate configuration may strongly impact on the performance of evolutionary algorithms. In this paper, we study the performance of three memetic algorithms for the quadratic assignment problem when their parameters are tuned either off-line or on-line. Off-line tuning selects a priori one configuration to be used throughout the whole run for all the instances to be tackled. On-line tuning selects the configuration during the solution process, adapting parameter settings on an instance-per-instance basis, and possibly to each phase of the search. The results suggest that off-line tuning achieves a better performance than on-line tuning.

Augmented reality for robots: Virtual sensing technology applied to a swarm of e-pucks

We present a novel technology that allows real robots to perceive an augmented reality environment through virtual sensors. Virtual sensors are a useful and desirable technology for research activities because they allow researchers to quickly and efficiently perform experiments that would otherwise be more expensive, or even impossible. In particular, augmented reality is useful (i) for prototyping and assessing the impact of new sensors before they are physically produced; and (ii) for developing and studying the behaviour of robots that should deal with phenomena that cannot be easily reproduced in a laboratory environment because, for example, they are dangerous (e.g., fire, radiations). We realised an augmented reality system for robots in which a simulator retrieves real-time data on the real environment through a multi-camera tracking system and delivers post-processed information to the robot swarm according to each robots sensing range. We illustrate the proposed virtual sensing technology through an experiment involving 15 e-pucks.

Multi-armed bandit formulation of the task partitioning problem in swarm robotics

Task partitioning is a way of organizing work consisting in the decomposition of a task into smaller sub-tasks that can be tackled separately. Task partitioning can be beneficial in terms of reduction of physical interference, increase of efficiency, higher parallelism, and exploitation of specialization. However, task partitioning also entails costs in terms of coordination efforts and overheads that can reduce its benefits. It is therefore important to decide when to make use of task partitioning. In this paper we show that such a decision can be formulated as a multi-armed bandit problem. This is advantageous since the theoretical properties of the multi-armed bandit problem are well understood and several algorithms have been proposed for tackling it. We carry out our study in simulation, using a swarm robotics foraging scenario as a testbed. We test an ad-hoc algorithm and two algorithms proposed in the literature for multi-armed bandit problems. The results confirm that the problem of selecting whether to partition a task can be formulated as a multi-armed bandit problem and tackled with existing algorithms.

Observing the effects of overdesign in the automatic design of control software for robot swarms

We present the results of an experiment in the automatic design of control software for robot swarms. We conceived the experiment to corroborate a hypothesis that we proposed in a previous publication: the reality gap problem bears strong resemblance to the generalization problem faced in supervised learning. In particular, thanks to this experiment we observe for the first time a phenomenon that we shall call overdesign. Overdesign is the automatic design counterpart of the well known overfitting problem encountered in machine learning. Past an optimal level of the design effort, the longer the design process is protracted, the better the performance of the swarm becomes in simulation and the worst in reality. Our results show that some sort of early stopping mechanism could be beneficial.