Abstract
Supernumerary multi-arm interfaces in Virtual Reality (VR) offer enhanced manipulation capabilities by adding virtual limbs beyond the user’s physical anatomy, but their utility depends heavily on effective control strategies. Zhou et al. (2025b) identified initial control strategies for such interfaces, but a systematic evaluation across diverse task types remains lacking. This study evaluates three control strategies—mapping-based, intent-driven, and haptic-guided—across four VR tasks (object sorting, tool operation, multi-target handling, and assembly). We tested 40 participants, measuring task performance (speed, accuracy) and subjective experience (comfort, perceived control). Results show that the intent-driven strategy outperforms others in complex tasks (assembly: 22.3% faster completion, p<0.01) but lags in simple tasks (sorting: 15.1% slower, p<0.05). The haptic-guided strategy achieves the highest comfort score (7.8/10 vs. 6.5/10 for mapping-based). These findings extend Zhou et al.’s (2025b) work, providing guidelines for strategy selection in supernumerary VR interfaces.
