Multi-Strategy Learning of Robotic Behaviours via Qualitative Reasoning

نویسنده

  • Timothy Wiley
چکیده

Introduction When given a task, an autonomous agent must plan a series of actions to perform in order to complete the goal. In robotics, planners face additional challenges as the domain is typically large (even infinite) continuous, noisy, and nondeterministic. Typically stochastic planning has been used to solve robotic control tasks. Particular success has been achieved with Model-Based Reinforcement Learning, such as Abbell et al (2010) which learnt a control policy for stable helicopter flight. Additionally, techniques such as Behavioral Cloning have been used to direct and speed up learning by providing knowledge from human experts (Michie, Bain, and Hayes-Michie 1990). The downside to such approaches is that the models and planners are highly specialised to a single control task. To change the control task, requires developing an entirely new planner. The research in my thesis focuses on the problem of specialisation in continuous, noisy and non-deterministic robotic domains. It builds on previous research in the area, specifically using the technique of Multi-Strategy Learning (Sammut and Yik 2010). I am using Qualitative Modelling and Qualitative Reasoning to provide the generality, from which specific, Quantitative controllers can be quickly learnt. The resulting system will be applied to the iRobot Negotiator Robotic Platform (Figure 1), on the domain task of traversing rough terrain types.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

A Multi-Strategy Architecture for On-Line Learning of Robotic Behaviours

A Multi-Strategy Architecture improves the efficiency of on-line learning of robotic behaviours by taking inspiration from approaches humans use for learning complex behaviours. The hybrid approach first learns the qualitative dynamics of a robotic system from which a symbolic planner constructs an approximate solution to a control problem by qualitatively reasoning over the discovered dynamics...

متن کامل

Qualitative Planning with Quantitative Constraints for Online Learning of Robotic Behaviours

This paper resolves previous problems in the Multi-Strategy architecture for online learning of robotic behaviours. The hybrid method includes a symbolic qualitative planner that constructs an approximate solution to a control problem. The approximate solution provides constraints for a numerical optimisation algorithm, which is used to refine the qualitative plan into an operational policy. In...

متن کامل

Planning with Qualitative Models for Robotic Domains

We aim to develop an online method for learning robot behaviours that requires only a small number of trials, making it practical for use on a real robot. A hybrid method includes a symbolic planner, using qualitative reasoning, that constructs an approximate solution to a control problem. This approximate solution provides constraints for a numerical optimisation algorithm, which is used to re...

متن کامل

A Planning and Learning Hierarchy using Qualitative Reasoning for the Online Acquisition of Robotic Behaviors

Trial-and-error learning is often needed to acquire a new skill. Humans can use domain knowledge to minimize the number of trials required. However, existing reinforcement learning systems are either incapable of reasoning about domain knowledge or use hard-coded domain knowledge. Thus, these systems are insufficient for the online learning of robotic skills. We present a hierarchical architect...

متن کامل

Improving Agent Performance for Multi-Resource Negotiation Using Learning Automata and Case-Based Reasoning

In electronic commerce markets, agents often should acquire multiple resources to fulfil a high-level task. In order to attain such resources they need to compete with each other. In multi-agent environments, in which competition is involved, negotiation would be an interaction between agents in order to reach an agreement on resource allocation and to be coordinated with each other. In recent ...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2013