Human Action Sequences Research Articles

Teaching and learning of robot tasks via observation of human performance

Within this paper, an approach for teaching a humanoid robot is presented that will enable the robot to learn typical tasks required in everyday household environments. Our approach, called Programming by Demonstration, which is implemented and successfully used in our institute to teach a robot system is presented. Firstly, we concentrate on an analysis of human actions and action sequences that can be identified when watching a human demonstrator. Secondly, sensor aid systems are introduced which augment the robot’s perception capabilities while watching a human’s demonstration and the robot’s execution of tasks respectively. The main focus is then layed on the knowledge representation in order to be able to abstract the problem solution strategies and to transfer them onto the robot system.

A methodology for modeling operator errors of commission in probabilistic risk assessment

Abstract This paper describes a methodology to incorporate operator errors of commission (EOCs) in nuclear power plant probabilistic risk assessments (PRAs). This is done by taking appropriate information from the plant PRA, operating procedures, information about plant configuration in terms of systems and functions, as well as physical and thermal-hydraulic information. This is combined with a set of performance influencing factors (PIFs), and results in an initial condition set which is fed into the primary tool for the methodology, called H uman I nteraction T ime LINE (HITLINE) to systematically generate sequences of human actions, including errors. Screening is applied to combinations of hardware failures, instrument failure, and PIFs to select the combinations that meet the criteria developed for the purpose. The criteria are based on the operator action or inaction causing a transition from one event tree (ET) branch to another. The strategy of utilizing mapping tables to accomplish all major steps of implementation decomposes the analysis into two separate parts. Values used in different scenario-dependent (error) likelihoods, and adjustors for these likelihoods to account for the influence of the scenario-independent PIFs, are separately assigned. While developing the HITLINE, the methodology then involves the use of mapping tables to generate a set of PIFs, given the relevant information about plant and emergency operating procedures (EOPs). This set is then used to select predetermined weights and adjustors to compute the final weight to be assigned to each branch at a single branching point. Quantification at each branching point is done through the multiple factors assigned to systematically assign the weights for different actions. Dependencies are carried from one branch point to another through the use of operator related variables such as operator diagnosis, and expectation about plant behavior. Size of HITLINE is managed by applying merging, truncation and termination rules at each time step. Similar end states with respect to the ET are terminated and their weights are combined. Combination is also applied among HITLINEs constructed for different initial condition sets for a given initiating event. Incorporating the results back into the ET either causes a reordering of sequence probabilities or including additional operator related top events to the ET. The methodology is demonstrated through a hypothetical example.

Learning by watching: extracting reusable task knowledge from visual observation of human performance

A novel task instruction method for future intelligent robots is presented, In our method, a robot learns reusable task plans by watching a human perform assembly tasks. Functional units and working algorithms for visual recognition and analysis of human action sequences are presented. The overall system is model based and integrated at the symbolic level. Temporal segmentation of a continuous task performance into meaningful units and identification of each operation is processed in real time by concurrent recognition processes under active attention control. Dependency among assembly operations in the recognized action sequence is analyzed, which results in a hierarchical task plan describing the higher level structure of the task. In another workspace with a different initial state, the system re-instantiates and executes the task plan to accomplish an equivalent goal. The effectiveness of our method is supported by experimental results with block assembly tasks.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

人間が実演して見せる作業の実時間視覚認識とそのロボット教示への応用

“Teaching by Showing” is a straightforward teaching method in which a human instructor shows an assembly task to a robot by simply performing it with his own hand, and the robot automatically generates a program by watching the example task. For such a method to be realized, visual recognition of human action sequences becomes a crucial issue. In this paper, an experimental system that demonstrates the teaching method is presented. The system observes example tasks by stereo video cameras and generates symbolic descriptions of action sequences, which are translated into manipulator command sequences. Actions are modeled as partial state changes over segmented time intervals. And the partial state descriptors constitute the environment model. Segmentation events are defined for actions involved in hand assembly tasks, which are detected in terms of time differences of visual features. Integrating these, an algorithm of action recognition is fully described. It automatically segments and classifies every action in a robust and efficient manner. An experiment revealed that the system can recognize each action correctly in real time, and it can carry out the instructed task successfully.

Knowledge-directed inference in believer

The BELIEVER theory is an attempt to specify an information processing systenn that constructs intentional interpretations of an observed sequence of human actions. A frame-based system, AIMDS, is used to define three domains: the physical world; the plan domain, where interpretations are constructed using plan structures composed from plan traits; and the psychological description of time actor. The system achieves a shift of representation from propositions about physical events to statements about beliefs and intentions of the actor by hypothesizing and attributing a.plan structure to the actor.