Project seeks to add object detection to wearables
MADISON, Wis. – Obstacle-detection technology, enabled by sensor fusion, is at the core of autonomous cars currently in development. The automotive industry has seen rapid progress in the technology. But what are the possibilities of applying the same sensor-fusion technology to wearable devices to aid the visually impaired or blind?
A team of researchers has banded together to take on this challenge. Their collaboration is part of a European project designed to develop “Integrated Smart Spatial Exploration System” called INSPEX.
The team seeks to develop “a portable and wearable, multi-sensor and low-power spatial-exploration and obstacle-detection system,” according to Leti, a technology research institute of CEA Tech. The goal is a white cane embedded with the INSPEX system -- with accessories that can provide 3D spatial audio feedback on obstacle location.
To make this work, the team must develop technologies that go several steps beyond the current state of the art in autonomous vehicles.
First, the requirements of the mission include devising an obstacle-detection system “for all conditions of weather and visibility.” Detection technology that works in smoke, dust, fog, heavy rain/snow, and darkness remains a tall order for today’s autonomous cars.
Second, the team needs to substantially reduce the weight and power of the portable system. Third, and perhaps most important, is reliability. Potential applications of the INSPEX system include drones and smart manufacturing (sssembly machines, security surveillance systems) in addition to guidance for the visually impaired and blind, according to Leti. “It will be used for real-time, 3D detection, location and warning of obstacles under all environmental conditions -- in indoor and outdoor environments with unknown stationary and mobile obstacles.”
The INSPEX system uses various range sensors such as LiDAR, UWB radar and MEMS ultrasound.
Suzanne Lesecq, director of research at Leti, also a coordinator for this project, told us that the project, funded by Horizon 2020 (the biggest EU Research and Innovation program), started on January 1. She’s shooting for a happy ending in 36 months (2019).
She promised that the team will have “a first proof of concept… validating our approach,” in December 2017. Among its expected elements will be “fusion of several range-sensing technologies on a low-power microcontroller to build a map of the user’s surroundings,” she said, referring to “sigma fusion” developed by Leti.
All weather conditions
Asked about all weather conditions the research team needs to meet, Lesecq said, “Yes, it is indeed very challenging. This is why we plan to use various range sensing technologies, including Lidar on chip, MEMS ultrasound and UWB RF radar brought by the partners.”
She added, “Our objective is for the INSPEX system to function under various weather conditions (e.g. rain, snow, sand) over a large temperature range (typ. -20°C to 40°C) but also in low visibility conditions (e.g. night, dust, smoke, fog). An environmental sensing module will help reconfigure the system depending on these environmental conditions.”
Considering that the INSPEX system needs to be installed inside a cane, what’s the target power consumption?
Continue reading page two on Embedded's sister site, EE Times: "Object-Detection becomes wearable."