Predictive action selector for generating meaningful robot behaviour from minimum amount of samples

Wieser, Erhard; Cheng, Gordon

doi:10.1109/devlrn.2014.6982969

Cited by 6 publications

(16 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Instead, it learns the sensory-motor associations from scratch, based on a simple constrained DOF exploration. In [5], we have grounded our PAS model in a cognitive science point of view. Here, we also ground our PAS model in a neurobiological point of view.…”

Section: Our Approachmentioning

confidence: 99%

“…Developmental studies such as [1], [2], [3], [4] point out that the acquisition of this visuo-proprioceptive mapping is the prerequisite for learning subsequent motor and cognitive skills, particularly related to hand-eye coordination. In order to solve this visuo-proprioceptive mapping problem on a real humanoid robot, we use our predictive action selector (PAS) proposed in [5], where our PAS controlled only the 2 DOF head of the humanoid robot NAO and facilitated the emergence of meaningful behaviour like object tracking and evading. However, it was not clear whether the original PAS [5] can be scaled up to deal with more than 2 DOF.…”

Section: Introduction a Learning The Sensory-motor Mapping As Prementioning

confidence: 99%

“…In order to solve this visuo-proprioceptive mapping problem on a real humanoid robot, we use our predictive action selector (PAS) proposed in [5], where our PAS controlled only the 2 DOF head of the humanoid robot NAO and facilitated the emergence of meaningful behaviour like object tracking and evading. However, it was not clear whether the original PAS [5] can be scaled up to deal with more than 2 DOF. In this paper, we explain the neurobiological grounding of Erhard Wieser and Gordon Cheng are with the Institute for Cognitive Systems, Technische Universität München, München, Germany, http://www.ics.ei.tum.de.…”

Section: Introduction a Learning The Sensory-motor Mapping As Prementioning

confidence: 99%

“…Our system starts with a constrained DOF exploration of the head joints (one DOF moving at a time, only once). During exploration, sensory-motor data is sampled and fed into our predictive action selector (PAS) [5] which we have improved in this paper. The PAS bootstraps head coordination (tracking behaviour and evading behaviour) from the training data.…”

Section: Introduction a Learning The Sensory-motor Mapping As Prementioning

confidence: 99%

“…By doing so, we emulate the pre-structuring of the human biological motor system in an early developmental stage, where the motions are constrained to simplify the learning [6], [1]. Compared to our original version [5], our improved PAS algorithm is more scalable and it has an increased accuracy in visual tracking. Compared to the related works, our proposed system is a neurobiologically-grounded approach for the progressive learning of sensory-motor maps.…”

Section: Introduction a Learning The Sensory-motor Mapping As Prementioning

confidence: 99%

See 4 more Smart Citations

Progressive learning of sensory-motor maps through spatiotemporal predictors

Wieser

Cheng

2016

2016 Joint IEEE International Conference on Development and Learning and Epigenetic Robotics (ICDL-EpiRob)

Self Cite

View full text Add to dashboard Cite

Abstract-Developmental robotics suggests that the forward and inverse kinematics should be learned through a sensorymotor mapping, instead of being programmed in advance. Motor babbling and goal babbling are two common approaches to generate training samples used to acquire a sensory-motor mapping. Motor babbling typically needs a considerable amount of training data and time to acquire a sufficient mapping, while goal babbling poses difficulties on how to select appropriate goals. In this paper, we propose a neurobiologically-inspired system to progressively learn a sensory-motor mapping bootstrapped from a simple constrained DOF exploration, which generates much less training data than motor babbling. Our proposed system is designed according to two neurobiologicallyinspired paradigms: spatiotemporal prediction and uniformity. The spatiotemporal prediction capability facilitates the acquisition of sensory-motor mappings with less amount of training data on the one hand, and facilitates robust behaviour on the other hand. The uniform system design structure is the foundation for building a scalable architecture for cognitive development. We use an improved version of our predictive action selector (PAS) as building block of our system. We validate a PAS on a 2 DOF robot head where the robot learns object tracking and evading. Then we validate a second PAS on a 5 DOF arm where it learns reaching.

show abstract

Section: Our Approachmentioning

confidence: 99%

Section: Introduction a Learning The Sensory-motor Mapping As Prementioning

confidence: 99%

Section: Introduction a Learning The Sensory-motor Mapping As Prementioning

confidence: 99%

Section: Introduction a Learning The Sensory-motor Mapping As Prementioning

confidence: 99%

Section: Introduction a Learning The Sensory-motor Mapping As Prementioning

confidence: 99%

See 3 more Smart Citations

Progressive learning of sensory-motor maps through spatiotemporal predictors

Wieser

Cheng

2016

2016 Joint IEEE International Conference on Development and Learning and Epigenetic Robotics (ICDL-EpiRob)

Self Cite

View full text Add to dashboard Cite

show abstract

EO-MTRNN: evolutionary optimization of hyperparameters for a neuro-inspired computational model of spatiotemporal learning

Wieser

Cheng

2020

Biol Cybern

Self Cite

View full text Add to dashboard Cite

For spatiotemporal learning with neural networks, hyperparameters are often set manually by a human expert. This is especially the case with multiple timescale networks that require a careful setting of the values of timescales in order to learn spatiotemporal data. However, this implies a cumbersome trial-and-error process until suitable parameters are found and it reduces the long-term autonomy of artificial agents, such as robots that are controlled by multiple timescale networks. To solve the problem, we propose the evolutionary optimized multiple timescale recurrent neural network (EO-MTRNN) that is inspired by the neural plasticity of the human cortex. Our proposed network uses a method of evolutionary optimization to adjust its timescales and to rewire itself in terms of number of neurons and synapses. Moreover, it does not require additional neural networks for pre-and postprocessing input-output data. We validate our EO-MTRNN by applying it to a proposed benchmark training dataset with single and multiple sequence training cases, as well as by applying it to sensory-motor data from a robot. We compare different configuration modes of the network, and we compare the learning performance between a network configuration with manually set hyperparameters and a configuration with automatically estimated hyperparameters. The results show that automatically estimated hyperparameters yield approximately 43% better performance than manually estimated ones, without overfitting the given teaching data. We also validate the generalization ability by successfully learning data that were not included in the hyperparameter estimation process.

show abstract