The problem of serial order: a neural network model of sequence learning and recall
Current research in natural language generation
Mechanism and process in animal behavior: models of animals, animals as models
SAB94 Proceedings of the third international conference on Simulation of adaptive behavior : from animals to animats 3: from animals to animats 3
Neural network dynamics for path planning and obstacle avoidance
Neural Networks
Eighteenth national conference on Artificial intelligence
The Effects of Frontal Lobe Lesions on Goal Achievement in the Water Jug Task
Journal of Cognitive Neuroscience
A self-organizing neural model of motor equivalent reaching and tool use by a multijoint arm
Journal of Cognitive Neuroscience
Spatial versus object working memory: Pet investigations
Journal of Cognitive Neuroscience
The focussed D* algorithm for real-time replanning
IJCAI'95 Proceedings of the 14th international joint conference on Artificial intelligence - Volume 2
Grounding action-selection in event-based anticipation
ECAL'07 Proceedings of the 9th European conference on Advances in artificial life
Neural network approaches to dynamic collision-free trajectorygeneration
IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics
Information Sciences: an International Journal
Adaptive Behavior - Animals, Animats, Software Agents, Robots, Adaptive Systems
Hi-index | 0.01 |
In order to create spatial plans in a complex and changing world, organisms need to rapidly adapt to novel configurations of obstacles that impede simple routes to goal acquisition. Some animals can mentally create successful multistep spatial plans in new visuo-spatial layouts that preclude direct, one-segment routes to goal acquisition. Lookahead multistep plans can, moreover, be fully developed before an animal executes any step in the plan. What neural computations suffice to yield preparatory multistep lookahead plans during spatial cognition of an obstructed two-dimensional scene? To address this question, we introduce a novel neuromorphic system for spatial lookahead planning in which a feasible sequence of actions is prepared before movement begins. The proposed system combines neurobiologically plausible mechanisms of recurrent shunting competitive networks, visuo-spatial diffusion, and inhibition-of-return. These processes iteratively prepare a multistep trajectory to the desired goal state in the presence of obstacles. The planned trajectory can be stored using a primacy gradient in a sequential working memory and enacted by a competitive queuing process. The proposed planning system is compared with prior planning models. Simulation results demonstrate system robustness to environmental variations. Notably, the model copes with many configurations of obstacles that lead other visuo-spatial planning models into selecting undesirable or infeasible routes. Our proposal is inspired by mechanisms of spatial attention and planning in primates. Accordingly, our simulation results are compared with neurophysiological and behavioral findings from relevant studies of spatial lookahead behavior.