Statistical multi-objective evaluation of human–robot interaction modes for collaborative robot control
Pratiksha Prakash Jawale, Shantipal S. Ohol, Kamalesh Arun SoratePurpose
Human–robot interaction (HRI) modalities have been extensively investigated in recent years; however, limited research has quantitatively compared different interaction modes within a unified experimental and statistical framework. The effectiveness of such interactions depends not only on the interface design but also on the operating conditions that influence both accuracy and efficiency.
Design/methodology/approach
This study presents a statistical analysis and multi-objective optimization framework to evaluate three modes of operation – voice, gesture and joystick – implemented on collaborative robot systems. Experiments were conducted using a replicated full factorial design with the speed scaling factor, operating distance and payload selected as control parameters. The positional error and task execution time were used as response variables to represent accuracy and efficiency, respectively. A general linear model-based ANOVA was performed to see the significance of the main factor and their interaction. Main effect and interaction analyses further studied to quantify the performance characteristics of each interaction interface. To minimize the positional error and execution time simultaneously, a multi-objective optimization framework based on response normalization and a comparative performance index (CPI) was developed.
Findings
The statistical analysis revealed that interaction modality, speed scaling factor and operating distance significantly influence system performance, while significant interaction effects are also observed among the operating parameters. Joystick-based interaction exhibited the highest overall CPI, followed by voice and gesture modes. The optimized condition was obtained for joystick operation at a speed scaling factor of 0.5, an operating distance of 100 mm and a payload of 4.5 kg, with a composite desirability of 0.9344.
Practical implications
This study provides practical insights for implementing multimodal control systems in collaborative robotic environments. By evaluating voice, gesture and joystick-based interaction using performance metrics such as positional error, task completion time and CPI, the research identifies suitable operating conditions for improved human–robot collaboration. The findings can assist industrial practitioners in selecting appropriate control modalities for tasks involving varying payloads, distances and speed factors. The proposed evaluation framework can support.
Originality/value
The proposed statistical evaluation and CPI-based optimization framework provide a practical methodology for selecting optimal HRI configurations for collaborative robotic applications.