Estimation of a Human-Maneuvered Target Incorporating Human Intention.

Area of Science:

• Robotics
• Control Systems
• Human-Robot Interaction

Background:

• Accurate motion state estimation is crucial for controlling targets maneuvered by humans.
• Conventional methods often struggle with the unpredictable nature of human intentions.
• Integrating human intention into estimation models can significantly improve performance.

Purpose of the Study:

• To propose a novel approach for estimating the motion state of human-maneuvered targets.
• To develop an intention-pattern model that associates recurring motion behaviors with human intentions.
• To enhance state estimation accuracy by incorporating this intention-pattern model.

Main Methods:

• Utilized an Interacting Multiple Model (IMM) estimation technique for inferring human intentions and extracting motion patterns.
• Constructed an intention-pattern model based on inferred intentions and extracted motions.
• Incorporated the intention-pattern model into state estimation using standard estimators like the Kalman filter.

Main Results:

• Demonstrated the effectiveness of the proposed approach in constructing the intention-pattern model.
• Showcased improved accuracy in estimating the mean and precision in updating the covariance of the target's state.
• Validated the approach through the state estimation of a human-maneuvered multirotor.

Conclusions:

• The proposed intention-pattern model significantly enhances motion state estimation for human-controlled targets.
• The method provides more accurate state estimation compared to conventional techniques lacking intention incorporation.
• This approach offers a promising direction for improving human-robot interaction and autonomous system control.