StairNet: visual recognition of stairs for human-robot locomotion

Abstract

Human-robot walking with prosthetic legs and exoskeletons, especially over complex terrains, such as stairs, remains a significant challenge. Egocentric vision has the unique potential to detect the walking environment prior to physical interactions, which can improve transitions to and from stairs. This motivated us to develop the StairNet initiative to support the development of new deep learning models for visual perception of real-world stair environments. In this study, we present a comprehensive overview of the StairNet initiative and key research to date. First, we summarize the development of our large-scale data set with over 515,000 manually labeled images. We then provide a summary and detailed comparison of the performances achieved with different algorithms (i.e., 2D and 3D CNN, hybrid CNN and LSTM, and ViT networks), training methods (i.e., supervised learning with and without temporal data, and semi-supervised learning with unlabeled images), and deployment methods (i.e., mobile and embedded computing), using the StairNet data set. Finally, we discuss the challenges and future directions. To date, our StairNet models have consistently achieved high classification accuracy (i.e., up to 98.8%) with different designs, offering trade-offs between model accuracy and size. When deployed on mobile devices with GPU and NPU accelerators, our deep learning models achieved inference speeds up to 2.8 ms. In comparison, when deployed on our custom-designed CPU-powered smart glasses, our models yielded slower inference speeds of 1.5 s, presenting a trade-off between human-centered design and performance. Overall, the results of numerous experiments presented herein provide consistent evidence that StairNet can be an effective platform to develop and study new deep learning models for visual perception of human-robot walking environments, with an emphasis on stair recognition. This research aims to support the development of next-generation vision-based control systems for robotic prosthetic legs, exoskeletons, and other mobility assistive technologies.

Keywords: Computer vision; Deep learning; Exoskeletons; Prosthetics; Wearable robotics.

Fig. 1

Inference and development pipelines for our baseline StairNet model [12] trained using supervised learning and single images. We developed this model as a reference and benchmark for the other deep learning models presented herein

Fig. 2

Model conversion and deployment pipeline for our mobile iOS application [13], which we developed to deploy and test our “Baseline Model” for on-device computing

Fig. 3

Inference and development pipelines for our temporal StairNet models [20] trained using supervised learning and sequential images. Unlike our previous models that used single image inputs, these temporal neural networks used sequential inputs

Fig. 4

Inference and development pipelines for our semi-supervised learning StairNet model [21] trained using labeled and unlabeled images. Unlike the aforementioned models, this model used large amounts of unlabeled data to minimize the number of required labelled images while still maintaining classification accuracy, therein improving training efficiency

Fig. 5

Model conversion and deployment pipeline for our smart glasses [22], which we developed to deploy and test our StairNet model for real-time embedded computing

Fig. 6

Inference and development pipelines for our smart glasses StairNet model trained using supervised learning and single images. Compared to our other models, the smart glasses performed stair recognition using a head-mounted camera and an embedded system