Limits...
Distributed recurrent neural forward models with synaptic adaptation and CPG-based control for complex behaviors of walking robots.

Dasgupta S, Goldschmidt D, Wörgötter F, Manoonpong P - Front Neurorobot (2015)

Bottom Line: Biological study has revealed that such complex behaviors are a result of a combination of biomechanics and neural mechanism thus representing the true nature of embodied interactions.Inspired by these findings, we present here, an artificial bio-inspired walking system which effectively combines biomechanics (in terms of the body and leg structures) with the underlying neural mechanisms.Furthermore, we demonstrate that the newly developed recurrent network based approach to online forward models outperforms the adaptive neuron forward models, which have hitherto been the state of the art, to model a subset of similar walking behaviors in walking robots.

View Article: PubMed Central - PubMed

Affiliation: Institute for Physics - Biophysics, George-August-University Göttingen, Germany ; Bernstein Center for Computational Neuroscience, George-August-University Göttingen, Germany ; Laboratory for Neural Computation and Adaptation, Riken Brain Science Institute Saitama, Japan.

ABSTRACT
Walking animals, like stick insects, cockroaches or ants, demonstrate a fascinating range of locomotive abilities and complex behaviors. The locomotive behaviors can consist of a variety of walking patterns along with adaptation that allow the animals to deal with changes in environmental conditions, like uneven terrains, gaps, obstacles etc. Biological study has revealed that such complex behaviors are a result of a combination of biomechanics and neural mechanism thus representing the true nature of embodied interactions. While the biomechanics helps maintain flexibility and sustain a variety of movements, the neural mechanisms generate movements while making appropriate predictions crucial for achieving adaptation. Such predictions or planning ahead can be achieved by way of internal models that are grounded in the overall behavior of the animal. Inspired by these findings, we present here, an artificial bio-inspired walking system which effectively combines biomechanics (in terms of the body and leg structures) with the underlying neural mechanisms. The neural mechanisms consist of (1) central pattern generator based control for generating basic rhythmic patterns and coordinated movements, (2) distributed (at each leg) recurrent neural network based adaptive forward models with efference copies as internal models for sensory predictions and instantaneous state estimations, and (3) searching and elevation control for adapting the movement of an individual leg to deal with different environmental conditions. Using simulations we show that this bio-inspired approach with adaptive internal models allows the walking robot to perform complex locomotive behaviors as observed in insects, including walking on undulated terrains, crossing large gaps, leg damage adaptations, as well as climbing over high obstacles. Furthermore, we demonstrate that the newly developed recurrent network based approach to online forward models outperforms the adaptive neuron forward models, which have hitherto been the state of the art, to model a subset of similar walking behaviors in walking robots.

No MeSH data available.


Related in: MedlinePlus

Snapshots showing the learned behavior during climbing over a high obstacle and climbing up a fleet of stairs. (A) AMOSII walked with the wave gait and approached a 15 cm high obstacle (1). It detected the obstacle using its range sensors installed at its front part. The low-pass filtered range sensory signals control the BJ to tilt upwards (2) and then back to its normal position (3). Due to the missing foot contact of the front legs, the BJ moved downwards to ensure stability (4). During climbing, middle and hind legs lowered downwards due to the occurrence of the accumulated errors, showing leg extension, to support the body. Finally, it successfully surmounted the high obstacle (5). For further details see the Supplementary Video 4(B) AMOSII climbed up a fleet of stairs (1–5) using the wave gait as well as the reactive BJ control. The climbing behavior is also similar to the one described in the case (A). For further details see Supplementary Video 5.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4585172&req=5

Figure 8: Snapshots showing the learned behavior during climbing over a high obstacle and climbing up a fleet of stairs. (A) AMOSII walked with the wave gait and approached a 15 cm high obstacle (1). It detected the obstacle using its range sensors installed at its front part. The low-pass filtered range sensory signals control the BJ to tilt upwards (2) and then back to its normal position (3). Due to the missing foot contact of the front legs, the BJ moved downwards to ensure stability (4). During climbing, middle and hind legs lowered downwards due to the occurrence of the accumulated errors, showing leg extension, to support the body. Finally, it successfully surmounted the high obstacle (5). For further details see the Supplementary Video 4(B) AMOSII climbed up a fleet of stairs (1–5) using the wave gait as well as the reactive BJ control. The climbing behavior is also similar to the one described in the case (A). For further details see Supplementary Video 5.

Mentions: Figure 8 shows that the reservoir forward model in combination with the neural locomotion control mechanisms, not only successfully generates gap crossing behavior of AMOS II and learns to walk on uneven terrain, but also allows it to climb over single and multiple obstacles (e.g., up a fleet of stairs). In all these cases, we directly used the accumulated errors for movement adaptation via the searching and elevation control mechanisms. For climbing, the reactive backbone joint control was also applied to the system (see Goldschmidt et al., 2014, for more details) and a slow wave gait walking pattern (see Figure 4A, right) was used.


Distributed recurrent neural forward models with synaptic adaptation and CPG-based control for complex behaviors of walking robots.

Dasgupta S, Goldschmidt D, Wörgötter F, Manoonpong P - Front Neurorobot (2015)

Snapshots showing the learned behavior during climbing over a high obstacle and climbing up a fleet of stairs. (A) AMOSII walked with the wave gait and approached a 15 cm high obstacle (1). It detected the obstacle using its range sensors installed at its front part. The low-pass filtered range sensory signals control the BJ to tilt upwards (2) and then back to its normal position (3). Due to the missing foot contact of the front legs, the BJ moved downwards to ensure stability (4). During climbing, middle and hind legs lowered downwards due to the occurrence of the accumulated errors, showing leg extension, to support the body. Finally, it successfully surmounted the high obstacle (5). For further details see the Supplementary Video 4(B) AMOSII climbed up a fleet of stairs (1–5) using the wave gait as well as the reactive BJ control. The climbing behavior is also similar to the one described in the case (A). For further details see Supplementary Video 5.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4585172&req=5

Figure 8: Snapshots showing the learned behavior during climbing over a high obstacle and climbing up a fleet of stairs. (A) AMOSII walked with the wave gait and approached a 15 cm high obstacle (1). It detected the obstacle using its range sensors installed at its front part. The low-pass filtered range sensory signals control the BJ to tilt upwards (2) and then back to its normal position (3). Due to the missing foot contact of the front legs, the BJ moved downwards to ensure stability (4). During climbing, middle and hind legs lowered downwards due to the occurrence of the accumulated errors, showing leg extension, to support the body. Finally, it successfully surmounted the high obstacle (5). For further details see the Supplementary Video 4(B) AMOSII climbed up a fleet of stairs (1–5) using the wave gait as well as the reactive BJ control. The climbing behavior is also similar to the one described in the case (A). For further details see Supplementary Video 5.
Mentions: Figure 8 shows that the reservoir forward model in combination with the neural locomotion control mechanisms, not only successfully generates gap crossing behavior of AMOS II and learns to walk on uneven terrain, but also allows it to climb over single and multiple obstacles (e.g., up a fleet of stairs). In all these cases, we directly used the accumulated errors for movement adaptation via the searching and elevation control mechanisms. For climbing, the reactive backbone joint control was also applied to the system (see Goldschmidt et al., 2014, for more details) and a slow wave gait walking pattern (see Figure 4A, right) was used.

Bottom Line: Biological study has revealed that such complex behaviors are a result of a combination of biomechanics and neural mechanism thus representing the true nature of embodied interactions.Inspired by these findings, we present here, an artificial bio-inspired walking system which effectively combines biomechanics (in terms of the body and leg structures) with the underlying neural mechanisms.Furthermore, we demonstrate that the newly developed recurrent network based approach to online forward models outperforms the adaptive neuron forward models, which have hitherto been the state of the art, to model a subset of similar walking behaviors in walking robots.

View Article: PubMed Central - PubMed

Affiliation: Institute for Physics - Biophysics, George-August-University Göttingen, Germany ; Bernstein Center for Computational Neuroscience, George-August-University Göttingen, Germany ; Laboratory for Neural Computation and Adaptation, Riken Brain Science Institute Saitama, Japan.

ABSTRACT
Walking animals, like stick insects, cockroaches or ants, demonstrate a fascinating range of locomotive abilities and complex behaviors. The locomotive behaviors can consist of a variety of walking patterns along with adaptation that allow the animals to deal with changes in environmental conditions, like uneven terrains, gaps, obstacles etc. Biological study has revealed that such complex behaviors are a result of a combination of biomechanics and neural mechanism thus representing the true nature of embodied interactions. While the biomechanics helps maintain flexibility and sustain a variety of movements, the neural mechanisms generate movements while making appropriate predictions crucial for achieving adaptation. Such predictions or planning ahead can be achieved by way of internal models that are grounded in the overall behavior of the animal. Inspired by these findings, we present here, an artificial bio-inspired walking system which effectively combines biomechanics (in terms of the body and leg structures) with the underlying neural mechanisms. The neural mechanisms consist of (1) central pattern generator based control for generating basic rhythmic patterns and coordinated movements, (2) distributed (at each leg) recurrent neural network based adaptive forward models with efference copies as internal models for sensory predictions and instantaneous state estimations, and (3) searching and elevation control for adapting the movement of an individual leg to deal with different environmental conditions. Using simulations we show that this bio-inspired approach with adaptive internal models allows the walking robot to perform complex locomotive behaviors as observed in insects, including walking on undulated terrains, crossing large gaps, leg damage adaptations, as well as climbing over high obstacles. Furthermore, we demonstrate that the newly developed recurrent network based approach to online forward models outperforms the adaptive neuron forward models, which have hitherto been the state of the art, to model a subset of similar walking behaviors in walking robots.

No MeSH data available.


Related in: MedlinePlus