Importance of Parameter Settings on the Benefits of Robot-to-Robot Learning in Evolutionary Robotics

Heinerman, Jacqueline; Haasdijk, Evert; Eiben, A. E.

doi:10.3389/frobt.2019.00010

Cited by 2 publications

(2 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Even more importantly, we have also shown that, as expected, social learning provides increased learning speed, but also yields increased performance when compared to individual learning [95]. This is due to the possibility of running several instances of individual learning algorithm with different meta-parameters values, as best values cannot be guessed before run-time.…”

Section: From One Agent To Anothersupporting

confidence: 64%

DREAM Architecture: a Developmental Approach to Open-Ended Learning in Robotics

Doncieux¹,

Bredèche²,

Goff³

et al. 2020

Preprint

View full text Add to dashboard Cite

Robots are still limited to controlled conditions, that the robot designer knows with enough details to endow the robot with the appropriate models or behaviors. Learning algorithms add some flexibility with the ability to discover the appropriate behavior given either some demonstrations or a reward to guide its exploration with a reinforcement learning algorithm. Reinforcement learning algorithms rely on the definition of state and action spaces that define reachable behaviors. Their adaptation capability critically depends on the representations of these spaces: small and discrete spaces result in fast learning while large and continuous spaces are challenging and either require a long training period or prevent the robot from converging to an appropriate behavior. Beside the operational cycle of policy execution and the learning cycle, which works at a slower time scale to acquire new policies, we introduce the redescription cycle, a third cycle working at an even slower time scale to generate or adapt the required representations to the robot, its environment and the task. We introduce the challenges raised by this cycle and we present DREAM (Deferred Restructuring of Experience in Autonomous Machines), a developmental cognitive architecture to bootstrap this redescription process stage by stage, build new state representations with appropriate motivations, and transfer the acquired knowledge across domains or tasks or even across robots. We describe results obtained so far with this approach and end up with a discussion of the questions it raises in Neuroscience.

show abstract

Section: From One Agent To Anothersupporting

confidence: 64%

DREAM Architecture: a Developmental Approach to Open-Ended Learning in Robotics

Doncieux¹,

Bredèche²,

Goff³

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…For a communication system to emerge, it is possible to use the evolutionary robotics (ER) approach (Bredeche et al, 2018). In ER the morphological and / or control structures are the result of the iterative pressure conducted by an artificial evolutionary process (Heinerman et al 2019;Nolfi, 1998). A common representation of control systems in ER are artificial neural networks (ANN) optimized by evolutionary algorithms (Alattas et al, 2019), such as Genetic Algorithms (GA) (Baldominos et al, 2020).…”

Section: Introductionmentioning

confidence: 99%

improvement of signal communication for a foraging task using evolutionary robotics

Aldana-Franco,

Montes-González,

Nolfi

2024

JART

View full text Add to dashboard Cite

Communication systems represent an evolutionary advantage for a group of robots solving coordinated tasks. In the field of evolutionary robotics, the emergence and establishment of communication are regulated by different variables. These systems are a tool produced by artificial evolution for the exchange of personal and environmental information. Since evolutionary processes involve multi-objective evolutionary methods, it is important to study all the mechanisms that affect the emergence and establishment of communication systems. One of these variables is the evolutionary advantage of emerging signals. In this article it is assumed that the signals appear because they are evolutionarily useful in solving a task. Emitter and receiver generate a process of conceptualization, which makes the signals associated with a meaning for the community. In this way an experiment is adapted to the FARSA simulator and the MARXBOT robot. This experiment consists of a group of robots that spend as much time as possible in a food zone and avoid a poisoned zone. Under normal conditions, robots tend to point to the food zone to attract the rest of the population. When the evolutionary importance of pointing out the food zone is reduced, the signals arise in different situations such as the presence of nearby objects and the poisoned zone. Communication is configured with the LED rings and linear cameras of the robots. In addition, an ethological method is adapted to quantify the behavioral effects of experimental manipulation.

show abstract

Importance of Parameter Settings on the Benefits of Robot-to-Robot Learning in Evolutionary Robotics

Cited by 2 publications

References 25 publications

DREAM Architecture: a Developmental Approach to Open-Ended Learning in Robotics

DREAM Architecture: a Developmental Approach to Open-Ended Learning in Robotics

improvement of signal communication for a foraging task using evolutionary robotics

Contact Info

Product

Resources

About