Integrating Human Inputs with Autonomous Behaviors on an Intelligent Wheelchair Platform
Peshala G Jayasekara
Outline • • • • • • •
Motivation Introduction Strategy Methodology Implementation Results Conclusion – Evaluation and Opinion
Motivation • The authors describe the development and assessment of a computer controlled wheelchair equipped with a suite of sensors and a novel interface, called the SMARTCHAIR. • Main focus - a shared control framework which allows the human operator to interact with the chair while it is performing an autonomous task.
Introduction (I) • Assistive Technology – Human-Robot Interaction (HRI) – Current Status of HRI • Supervisory control • Less Trust
• We need HRI to be similar to two individuals working together, where each one can predict the other’s behavior
Introduction (II) • Research on computer-controlled chairs – Wheel-chairs for disabled
• Approaches – Total control via Joystick – User’s desire via commands, and the chair moves autonomously – Motion through following • eg. Wall-follow
• Main Drawback – Such HRI systems tend to grow anxiety, frustration and distrust on users
Strategy • Sync together three diverse and at times contradictory goals in motion planning • Deliberative • Reactive • User-initiated
- maps, user commands - stimuli from environment - unpredicted user behavior
Methodology (I) • Deliberative motion plan - Potential Field Method
Methodology (II) • Deliberative Plan with Reactive Behaviors
• Deliberative Plan, Reactive Behaviors, and Human Inputs – Hierarchical, prioritized shared control framework
Implementation • HRI – vision based – Cameras, Projector
Implementation • HRI – vision based – Cameras, Projector
• Obstacle Avoidance – Laser scanner, IR proximity sensors
• Dead reckoning – Encoders, overhead markers
Results (I)
Sample trajectory for autonomous mode The trajectory is a result of the deliberative controller (potential field controller) and local, reactive behaviors.
Sample trajectory (red) for manual mode The blue line segments represent the activation of obstacle avoidance and modification of the human input.
Results (II)
A sample trajectory using a deliberative plan, combined with user input (semiautonomous mode): (c) the deliberative path taken, (d) the user’s input, which is consistent with the deliberative plan (e) when the obstacle avoidance behavior is activated
Quantitative and qualitative analysis
Conclusion • Assistive Technology • A SmartChair with shared control strategy to incorporate human inputs in motion planning • How the potential field method was modified • Affects on frustration and sense of control in HRI
Evaluation and Opinion • Camera – projector interface is creative • Effort to reduce frustration factor • The team has spent time not only to produce a “smart” chair but also to get the opinion of the end users. • No outdoor testing results • Chair is not eye-catching
Thank you
Questions??
Equations