Researchers at The University of Tokyo have revealed a new 3D motion sensing technique which uses multiplexed lasers to detect an object’s movement without prior knowledge of its shape or texture.
The team, based at the university’s Ishikawa Watanabe Lab, explained that current object motion is recorded and used in various applications, with a diverse range of sensed objects and devices. However, according to the scientists, these sensing methods are inflexible and have ‘low universality’ as they require equipment to be installed on the target, such as accelerometers and gyro sensors. They added that even camera-based sensing systems are limited to the object shape and texture and demand high calculation costs. Notably, camera methods ‘inevitably fail’ should the target object have no distinctive surface texture or structural features.
The proposed laser solution is able to overcome these challenges, allowing for non-contact, high-speed, deterministic recordings of the velocity of a moving object without previous details of its properties. The method can therefore be applied to any ‘unconstrained, unspecified’ object.
The novel sensing system consists of a laser range finder, a laser Doppler velocimeter, a beam controller, and is managed by ‘robust 3D motion calculation’ algorithms. The movement of the object is compiled from physical data, including the distance and speed of the object measured at the laser irradiation points.
Using this system, the researchers developed a number of applications, such as motion capture – manipulating virtual characters using real-life objects. They also experimented using everyday objects as video game user interfaces, such as employing a cardboard cut-out as a controller for a racing game. Further research combined the system with a computer vision field and a depth sensor to enable high-resolution 3D scanning of moving everyday objects.
Object sensing and recognition is a hot subject, not only for gaming and animation use cases, but is growing in popularity among security and surveillance research projects. A recent method for protecting privacy of individuals caught on camera, has been proposed by the University of Zagreb, which is able to sense a target’s movement and replace it with a CGI equivalent.