New Approach to System Safety of human- multi-robot mobile system control with STPA and FTA

Abstract

Autonomous Mobile multi-robots are among the most complex systems in their control. Especially when those robots navigate in hazardous and dynamic environments such as chemical analysis laboratories which include dangerous and harmful products (poisonous, flammable, explosive ...). This study deals the safety problem in a robotic analysis laboratory and investigates the possibility to use those autonomous multi-robots in such environments with the presence of human workers without serious hazards. We used a systems-theoretic hazard analysis technique (STPA) in addition to fault tree analysis to identify the potential safety hazard scenarios, their causal factors and we conclude by a set of recommendations. Keywords: Hazard Analysis, STPA, FTA, Collaborative multi-mobile robots.

STPA for Learning-Enabled Systems: A Survey and A New Practice

“…Autonomous ships and drones are also studied in [58], [49], [27], [61], [5] and [16], respectively. Collaborative robots [11], [3], [14] and Autonomous Mobile Multi-robots [9], [10], [14], [12] are also active areas of applying STPA. Moreover, STPA-priv was initially implemented for smart televisions [53] before being extended to E-health [39], recognising that privacy is a vital concern for humans.…”

“…STPA appears to be effective in analysing such complex CPSs, thanks to its holistic and systematic approach, emphasising causality for early hazard identification. c) Modification of the Method: While STPA can be directly applied on LESs, due to the intricacy of LESs, many studies advocate combining STPA with other safety analysis methods, e.g., the aforementioned STPA-sec and STPA-priv, the combined use of STPA and Causal Analysis based on System Theory (CAST) [2], a hybrid method for assessing system reliability combines STPA and Success Likelihood Index Method (SLIM) [5], and the combination of STPA with Uncontrolled Flows of Information and Energy (UFoI-E) [27], Bow Tie methodology (Bowtie) [11], Fault tree analysis (FTA) [9] and multilevel run-time monitors [24]. It is also feasible to integrate STPA with hierarchical modelling approaches for intricate systems [50], [31].…”

“…8 Data augmentation diversifies and increases training data to improve the model's ability to generalize and handle real-world variations. 9 The ε-greedy approach is a strategy that enables the agent to explore different actions while also exploiting the currently known optimal actions. 10 Replaying past experiences during training helps the model learn from diverse sets of data, enhancing efficiency and stability.…”