Efficient Indoor Positioning with Visual Experiences via Lifelong Learning

Abstract

Abstract-Positioning with visual sensors in indoor environments has many advantages: it doesn't require infrastructure or accurate maps, and is more robust and accurate than other modalities such as WiFi. However, one of the biggest hurdles that prevents its practical application on mobile devices is the time-consuming visual processing pipeline. To overcome this problem, this paper proposes a novel lifelong learning approach to enable efficient and real-time visual positioning. We explore the fact that when following a previous visual experience for multiple times, one could gradually discover clues on how to traverse it with much less effort, e.g. which parts of the scene are more informative, and what kind of visual elements we should expect. Such second-order information is recorded as parameters, which provide key insights of the context and empower our system to dynamically optimise itself to stay localised with minimum cost. We implement the proposed approach on an array of mobile and wearable devices, and evaluate its performance in two indoor settings. Experimental results show our approach can reduce the visual processing time up to two orders of magnitude, while achieving sub-metre positioning accuracy.

A smartphone-based zero-effort method for mitigating epidemic propagation

“…Recognizing the importance of contact tracing, many studies on contact tracing systems have been developed [32]. Some commercial solutions conduct contact tracing with GPS [33], RFID [34], ultra-wideband [35], BLE [36][37][38], Wi-Fi [31,39], cellular [40][41][42], vision [43], and other technologies. To sense mobile social interactions, Banerjee et al [44] proposed virtual compass, which effectively perceives the mutual distance between users, but cannot obtain direction information.…”

A Smartphone-based Zero-Effort Method for Mitigating Epidemic Propagation

“…Recognizing the importance of contact tracing, many studies on contact tracing systems have been developed [35]. Some commercial solutions conduct contact tracing with GPS [36], RFID [37], ultrawideband [38], BLE [39]- [41], Wi-Fi [34], [42], cellular [43]- [45], vision [46] and other technologies. To sense mobile social interactions, Banerjee et al [47] proposed virtual compass , which effectively perceives the mutual distance between users, but cannot obtain direction information.…”

Vision-Based 2D Navigation of Unmanned Aerial Vehicles in Riverine Environments with Imitation Learning

“…In the repeating pass, the robot will localize and plan a path to track the pose of keyframes accordingly. The VT&R has been tested successful in indoor environments [56] and outdoor long-range rover autonomy by using stereo camera [12] and monocular vision [8], and can also deal with the seasonal changes of the environments [25]. However, the VT&R requires the robots to repeat a previously traversed route since it utilizes the maps built in the teaching phase.…”