Implementing accurate run-time and resource- efficient risk predictions for human-robot interaction is an open challenge as it requires executing long and resource-intensive tasks. This paper addresses this challenge by presenting a methodology to avoid unwanted outcomes in the context of human-robot interaction, with the goal of minimising the cost of prediction. It is based on a two-phase approach. First, the occurrence of risky situations is monitored by a ”light” assertion mining. The miner continuously analyses the execution traces that describe the behaviours of humans and robots, searching for risk conditions expressed through logic formulas. When any of these is detected, the miner automatically extracts the causes that led to the risk situation. At that point, it activates the more accurate and ”heavier” predictor to monitor the risk of collision. This leads to sensible improvements in the prediction and resource- saving of the overall monitoring system. Experimental results have been conducted on an industrial case study implementing a smart manufacturing line with a Kuka LBR IIWA R820 robot.

Risk Assessment and Prediction in Human-Robot Interaction Through Assertion Mining and Pose Estimation

Michele Boldo;Nicola Bombieri;Mirco De Marchi;Luca Geretti;Samuele Germiniani;Graziano Pravadelli
2022

Abstract

Implementing accurate run-time and resource- efficient risk predictions for human-robot interaction is an open challenge as it requires executing long and resource-intensive tasks. This paper addresses this challenge by presenting a methodology to avoid unwanted outcomes in the context of human-robot interaction, with the goal of minimising the cost of prediction. It is based on a two-phase approach. First, the occurrence of risky situations is monitored by a ”light” assertion mining. The miner continuously analyses the execution traces that describe the behaviours of humans and robots, searching for risk conditions expressed through logic formulas. When any of these is detected, the miner automatically extracts the causes that led to the risk situation. At that point, it activates the more accurate and ”heavier” predictor to monitor the risk of collision. This leads to sensible improvements in the prediction and resource- saving of the overall monitoring system. Experimental results have been conducted on an industrial case study implementing a smart manufacturing line with a Kuka LBR IIWA R820 robot.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11562/1074267
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact