Brenan J. McCarragher

Skill acquisition from human demonstration using a hidden Markov model

A new approach to skill acquisition in assembly is proposed. An assembly skill is represented by a hybrid dynamic system where a discrete event controller models the skill at the task level. The output of the discrete event controller provides the reference commands for the underlying robot controller. This structure is naturally encoded by a hidden Markov model (HMM). The HMM parameters are obtained by training on sensory data from human demonstrations of the skill. Currently, assembly tasks have to be performed by human operators or by robots using expensive fixtures. Our approach transfers the assembly skill from an expert human operator to the robot, thus making it possible for a robot to perform assembly tasks without the use of expensive fixtures.

A discrete event approach to the control of robotic assembly tasks

An approach to process modeling, task synthesis, and motion control for robotic assembly is presented. Assembly is modeled as a discrete-event dynamic system using Petri nets and incorporating both discrete and continuous aspects of the process. To accomplish a desired trajectory, a discrete event controller is developed. The controller issues velocity commands that direct the system toward the next desired contact state, while maintaining currently desired contacts and avoiding unwanted transitions. Experimental results are given for a dual peg-in-the-hole example. The results not only demonstrate highly successful insertion along the optimal trajectory, but also demonstrate the ability to detect, recognize, and recover from errors and unwanted situations.<<ETX>>

Task primitives for the discrete event modeling and control of 6-DOF assembly tasks

A novel approach to the modeling and control of six degree-of-freedom (DOF) assembly tasks using polygonal pacts is presented. The assembly process is modeled as a discrete event system. Three key features of the assembly process are shown to be essential for a comprehensive understanding and effective control of the assembly task. It is also shown that discrete event modeling highlights these key aspects of the assembly process. Next the assembly process is broken down into its task primitives, or fundamental states of contact. It is shown that with only two primitives, the entire assembly process can be represented. These two task primitives are a surface-vertex contact and an edge-edge contact. The significance of this development is demonstrated. Petri nets are then used to give the discrete event model its necessary mathematical description for robotic control applications. The control methodology, again based on the two primitives, establishes a powerful framework which is robust and flexible in the control of assembly. Finally, simple experiments are discussed in which a successful 6-DOF assembly was completed.

Force sensing from human demonstration using a hybrid dynamical model and qualitative reasoning

This paper presents an analysis of force data that was generated by a human demonstration of a complex and asymmetrical insertion task. The human was blindfolded. And only had force sensors (i.e. fingers) available. The process is modelled using hybrid dynamics which highlighted the changes in contact as the essential parts of the assembly process. Qualitative reasoning is then applied to the force signals to identify those portions of the force signals that corresponded to either a gain or a loss of contact. The emphasis of the paper is to understand the force signals generated by the human subject. It is shown that qualitative reasoning proves to be a simple, efficient and effective means to understand the force signal by identifying the changes in contact state. Qualitative reasoning is also shown to be suitable for automated industrial assembly.<<ETX>>

Human integration into robot control utilising potential fields

Human integration into discrete event control systems is considered by utilising potential fields for control synthesis. Human integration into a control system is of importance in. Many situations, including those in which limited sensor information about the environment or the task is available. The framework presented allows the sharing of human commands with commands from an automated control system. Potential fields are being used as a tool to generate velocity commands from an autonomous task level controller as well as allowing the human to interact. The potential fields are also utilised to constrain human input such that human input error is minimised. The ideas presented here are supported by experiments.

Dynamic sensor selection for robotic systems

A new technique for selecting, in real time, different sensing techniques for a robotic system has been developed. The proposed method is based on stochastic dynamic programming, which provides an effective solution to multi-stage decision problems. At each stage in the decision process a sensor selection controller has the option of consulting a new process monitoring technique to improve the knowledge of the task or terminating the decision process without any further information gathering. The sensor selection controller has been successfully implemented for the real-time control of a planar robotic assembly task in a discrete event control framework. One of the monitoring methods used is based on hidden Markov models, where the average recognition rate was 87%. The rate of 87% was chosen to show the effectiveness of the dynamic sensor selection method. The experiments show that the method performs better than any individual process monitor. A successful event recognition rate of 97% with an average CPU time of 0.38 seconds is achieved when two force monitors and one position monitor are available to the sensor selection controller.

Petri net modelling for robotic assembly and trajectory planning

A new approach to process modelling, task synthesis, motion control and trajectory planning for robotic assembly is presented. Assembly is modelled as a discrete event dynamic system using Petri nets, incorporating both discrete and continuous aspects of the process. A process monitor based on recognizing contact state transitions is presented. A discrete event controller is developed. The controller issues velocity commands that direct the system toward the next desired contact state, while maintaining currently desired contacts and avoiding unwanted transitions. A novel means of trajectory planning which incorporates the systems ability to both monitor and control the process is given. Experimental results are given for a dual peg-in-the-hole example. The experimental results not only demonstrate highly successful insertion along the desired trajectory, but also demonstrate the ability to detect, recognize, and recover from errors and unwanted situations. >

Modeling and constraining human interactions in shared control utilizing a discrete event framework

Human integration is essential in systems where autonomous control alone would not be successful. In this paper, we present a framework for integrating a human supervisor into an otherwise autonomous control system. To facilitate integration, discrete event systems theory is adopted to model human interactions. It is via this interaction model that human commands can be combined with commands from an automated control system. For control synthesis, a method based on constraints is being used to generate velocity commands from the autonomous task level controller. The constraints are also utilized to limit human input so that erroneous human input is minimized. The methods are demonstrated by experiments.

Frequency-domain force measurements for discrete event contact recognition

Discrete event recognition based on force measurements in the frequency-domain, is presented. The force signals arise from interaction between the workpiece and the environment in a planar assembly task. The discrete events are modeled as hidden Markov models (HMMs), where the models are trained off-line with the Baum-Welch re-estimation algorithm. After the HMMs have been trained, we use them online in a robotic system to recognise discrete events as they occur. Event recognition with an accuracy as high as 98% was accomplished in 0.5-0.6 s with a relatively small training set.

Geometric constraint identification and mapping for mobile robots

Abstract A new method of map building for mobile robots is presented. Recent developments have focused on grid-based mapping methods which suffer from the drawback of their size, requiring a great deal of memory and prohibiting the use of many path-planning algorithms. In contrast, geometric maps provide a compact alternative which facilitates path-planning. We propose a new method which identifies geometric models of the constraints imposed upon the robot by the environment. A rigorous approach is taken to the process of constraint identification, which is cast as a minimisation problem. A number of primitive geometric objects are used for constraint modelling including line segments, arc segments, cubic segments and, for three degree of freedom systems, polygonal planar patches. A number of operations are also defined which integrate new sensor readings into the existing model. Simulation results are presented for two and three degree of freedom systems, demonstrating the effectiveness of the constraint identification process. A comparative study is also presented which gives guidelines for the proper selection of primitives and operations.