Siamese Object Tracking for Vision-Based UAM Approaching with Pairwise Scale-Channel Attention

Abstract

Visual approaching to the target object is crucial to the subsequent manipulating of the unmanned aerial manipulator (UAM). Although the manipulating methods have been widely studied, the vision-based UAM approaching generally lacks efficient design. The key to the visual UAM approaching lies in object tracking, while current approaching generally relies on costly model-based methods. Besides, UAM approaching often confronts more severe object scale variation issues, which makes it inappropriate to directly employ state-of-the-art model-free Siamese-based methods from the object tracking field. To address the above problems, this work proposes a novel Siamese network with pairwise scale-channel attention (SiamPSA) for vision-based UAM approaching. Specifically, SiamPSA consists of a scale attention network (SAN) and a scale-aware anchor proposal network (SA-APN). SAN acquires valuable scale information for feature processing, while SA-APN mainly attaches scale-awareness to anchor proposing. Moreover, a new tracking benchmark for UAM approaching, namely UAMT100, is recorded with 35K frames on a flying UAM platform for evaluation. Exhaustive experiments on the benchmark and real-world tests validate the efficiency and practicality of SiamPSA with a promising speed. Both the code and UAMT100 benchmark are now available at https://github.com/vision4robotics/SiamPSA.

Publication
In Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Kyoto, Japan, pp. 1-8, 2022.

SiamPSA_workflow An overview of the proposed Siamese tracking with pairwise scale-channel attention (SiamPSA) for UAM approaching.

Avatar
Guangze Zheng
PhD Candidate in Computer Science at HKU, China