Learned models for continuous planning
نویسندگان
چکیده
We are interested in the nature of activity { structured behavior of nontrivial duration { in intelligent agents. We believe that the development of activity is a continual process in which simpler activities are composed, via planning, to form more sophisticated ones in a hierarchical fashion. The success or failure of a planner depends on its models of the environment, and its ability to implement its plans in the world. We describe an approach to generating dynamical models of activity from real-world experiences and explain how they can be applied towards planning in a continuous state space.
منابع مشابه
Model-Based Planning with Discrete and Continuous Actions
Action planning using learned and differentiable forward models of the world is a general approach which has a number of desirable properties, including improved sample complexity over modelfree RL methods, reuse of learned models across different tasks, and the ability to perform efficient gradient-based optimization in continuous action spaces. However, this approach does not apply straightfo...
متن کاملNonlinear Hybrid Planning with Deep Net Learned Transition Models and Mixed-Integer Linear Programming
In many real-world hybrid (mixed discrete continuous) planning problems such as Reservoir Control, Heating, Ventilation and Air Conditioning (HVAC), and Navigation, it is difficult to obtain a model of the complex nonlinear dynamics that govern state evolution. However, the ubiquity of modern sensors allow us to collect large quantities of data from each of these complex systems and build accur...
متن کاملLearning Grounded Relational Symbols from Continuous Data for Abstract Reasoning
Learning from experience how to manipulate an environment in a goal-directed manner is one of the central challenges in research on autonomous robots. In the case of object manipulation, efficient learning and planning should exploit the underlying relational structure of manipulation problems and combine geometric state descriptions with abstract symbolic representations. When appropriate symb...
متن کاملTowards Supply Chain Planning Integration: Uncertainty Analysis Using Fuzzy Mathematical Programming Approach in a Plastic Forming Company
Affected by globalization and increased complexity, supply chain managers have learned about the importance of Sales and Operations Planning (S&OP). However, in large scale supply chains, S&OP has received little attention, by both academics and practitioners. The purpose of this manuscript is to investigate the advantages of S&OP process using a mathematical modeling approach in a large scale ...
متن کاملActive model learning and diverse action sampling for task and motion planning
The objective of this work is to augment the basic abilities of a robot by learning to use new sensorimotor primitives to enable the solution of complex long-horizon problems. Solving long-horizon problems in complex domains requires flexible generative planning that can combine primitive abilities in novel combinations to solve problems as they arise in the world. In order to plan to combine p...
متن کامل