notesum.ai

Published at November 27

SCoTT: Wireless-Aware Path Planning with Vision Language Models and Strategic Chains-of-Thought

cs.LG

cs.AI

cs.RO

cs.SY

eess.SY

Released Date: November 27, 2024

Authors: Aladin Djuhera¹, Vlad C. Andrei¹, Amin Seffo¹, Holger Boche¹, Walid Saad²

Aff.: ¹Technical University of Munich, Munich, Germany; ²Virginia Tech, Arlington, VA, USA

Arxiv: http://arxiv.org/abs/2411.18212v1

Refer to caption

		Path 1: Across the Room	Path 2: Wall to Wall	Path 3: Extreme Case
Classical A*	Average Path Gain [norm.]	0.34	0.06	0.10
	Total Path Length [m]	7.43	7.07	2.90
	Time [s]	2.23	2.24	1.01
Naïve A*	Average Path Gain [norm.]	0.46	0.31	0.22
	Total Path Length [m]	7.77	9.30	3.83
	Time [s]	4.3	5.32	3.53
DP-WA*	Average Path Gain [norm.]	0.75	0.43	0.69
	Total Path Length [m]	9.15	10.21	9.63
	Time [s]	76.04	74.86	76.97
SCoTT	Average Path Gain [norm.]	0.72	0.41	0.62
	Total Path Length [m]	8.64	9.32	8.22
	Time [s]	–	–	–
SCoTT- DP-WA*	Average Path Gain [norm.]	0.73	0.43	0.66
	Total Path Length [m]	9.38	10.21	9.40
	Time [s]	30.01	29.17	33.76