Announcement

Collapse
No announcement yet.

Navigation Tip

Collapse

To find what you need, select a category and then a project from the left sidebar.

Agents

Collapse

In this category, you'll find a variety of agents developed for a wide range of different tasks and environments. You will find a full description of the agent, its capabilities and problem solving approach as well, as a download link. If the agent requires an environment, a link will be provided. We'd highly encourage you to submit your own agents for use by the greater Soar community. To do so, you can send your zipped up submission with a full description (try to include all of the type of information we include on each download page) to mazina@umich.edu with the subject "Soar Agent Submission".

  •  
  • Filter
  • Time
  • Show
Clear All
new posts

  • Water Jug (Reinforcement Learning)

    Water Jug (Reinforcement Learning)

    This agent demonstrates how to solve the water jug problem using reinforcement learning. It is a modification of the Simple Water Jug Agent. Two main changes are made: templates are used to generate the full space of possible moves that the agent can perform and rewards are set based on the problem solution.

    Soar capabilities
    • Reinforcement learning
    Download LinksExternal Environment
    • None.
    Default Rules
    • None.
    Associated Publications
    • The
    ...
    See more | Go to post

  • Blocks-World (Reinforcement Learning)

    Blocks-World (Reinforcement Learning)

    This agent contains a version of blocks-world that uses reinforcement learning for move-block, which moves a block to a destination.

    Soar capabilities
    • Reinforcement Learning
    Download LinksExternal Environment
    • None.
    Default Rules
    • None.
    Associated PublicationsDeveloper
    • John Laird
    Soar Versions
    • Soar 8,9
    Project Type
    • VisualSoar
    See more | Go to post

  • Blocks-World (Subgoaling with RL)

    Blocks-World (Subgoaling with RL)

    agent incorporates both operator subgoaling/means ends analysis with reinforcement learning. All search control knowledge (operator evaluation rules) are removed from blocks-world-operator-subgoaling and instead there are RL rules supplemented with rules to compute reward, both in the top state and the substate. Implemented for four blocks.

    Soar capabilities
    • Subgoaling with means-ends analysis
    • Reinforcement learning
    Download LinksExternal En...
    See more | Go to post

  • Water Jug (Look-Ahead with State Evaluation)

    Water Jug (Look-Ahead with State Evaluation)

    This agent is a modification of the Water Jug Simple Agent that demonstrates using a tie impasse to subgoal and evaluate operators. This is an excellent demonstration of look-ahead search and how the selection default rules work.

    Unlike the Water Jug Look-Ahead Agent, the tie agent has knowledge that prefers certain moves over others, for example preferring a pour after a fill operator, so it is able to solve the problem much more efficiently.

    Soar capabilities
    • Look-ahead
    ...
    See more | Go to post

  • Water Jug (Hierarchical Task Decomposition)

    Water Jug (Hierarchical Task Decomposition)

    This project contains a version of the water jug problem that is formulated for hierarchical task decomposition. It involves two levels of problem spaces. The top level has three operators: fill, empty or pour. The next level consists of three operators: pick-up, fill-jug and put-down and they arise in an operator no-change for the super state.

    Note: This agent works with chunking, which compiles the actions in the substates into rules that apply at the top-state. Use the command "learn...
    See more | Go to post

  • Water Jug (Look-Ahead)

    Water Jug (Look-Ahead)

    This agent is a modification of the Water Jug Simple Agent that demonstrates using a tie impasse to subgoal and evaluate operators. This is an excellent demonstration of look-ahead search and how the selection default rules work.

    Unlike the Water Jug Look-Ahead Agent with State Evaluation , the look-ahead agent has no knowledge that prefers certain moves over others, so it does exhaustive search which takes far more decision cycles.

    Soar capabilities
    • Look-ahead subgoaling
    • Can
    ...
    See more | Go to post

  • Taxi (Hierarchical Reinforcement Learning)

    Taxi (Hierarchical Reinforcement Learning)

    This agent simulates an omniscient taxi driver that uses reinforcement learning and hierarchical task decomposition to improve its performance over runs.

    Soar capabilities
    • Hierarchical task decomposition
    • Reinforcement learning
    Download Links
    • The agent is included in the Taxi download.
    External EnvironmentDefault Rules
    • None.
    Associated Publications...
    See more | Go to post

  • Taxi (Reinforcement Learning)

    Taxi (Reinforcement Learning)

    This agent simulates an omniscient taxi driver that uses reinforcement learning to improve its performance over runs.

    Soar capabilities
    • Reinforcement Learning
    Download Links
    • The agent is included in the Taxi download.
    External EnvironmentDefault Rules
    • None.
    Associated PublicationsDevel...
    See more | Go to post

  • Blocks-World (Hierarchical Look-Ahead)

    Blocks-World (Hierarchical Look-Ahead)

    This project augments the Blocks-World Hierarchical Agent with look-ahead state evaluation. The description of the original agent applies to this one. The main difference is that look-ahead is performed in the middle of the three problem spaces that it uses.

    Soar capabilities
    • Hierarchical task composition via subgoaling
    • Look-ahead subgoaling
    • Internally simulates external environment including an i/o link
    • Can learn procedural knowledge (enable with 'learn always')
    Download Li...
    See more | Go to post

  • Blocks-World (Hierarchical)

    Blocks-World (Hierarchical)

    This project contains a version of blocks world that is formulated for hierarchical task decomposition. It involves three levels of problem spaces. There is sufficient evaluation knowledge so that there is no search/uncertainty at every level. The top level has a single operator: move-block, which moves a block (moving-block) to a destination. The destination can be the top of another block or the table.

    The next level consists of two operators: pick-up and put-down and they arise in...
    See more | Go to post
There are no articles in this category.
kusadasi escort
alanya escort
free sex porn hd porno
huluhub.com
ankara escort pendik escort
umraniye escort
Kayseri escort Eskisehir escort Eskisehir escort
replica rolex
Working...
X