In Yamakawa laboratory, by connecting various sensors on the network, centered on high-speed vision, we construct sensor network systems to quickly and comprehensively recognize the real world. Then we also feedback to actuation systems (robots etc.) in real time. We aim to develop high-speed intelligent systems that realize dynamic interaction. Currently, we are pursuing developments and research in the fields below. Among sensing networks, we have realized high-speed sensor networks and ITS applications. Meanwhile, we have realized actuation systems for high-speed, dynamic manipulation and human-robot interaction.

・High-speed sensor networks
In research on high-speed camera networks, we have developed a measurement system that can capture and process 1,000 images per second and are studying how to detect and stably track multiple objects in a large area. The features of high speed and networking make it possible to observe the dynamic motion of objects with seamless spatiotemporal information. Our research is expected to be applied to various IoT systems including security systems.

・ITS applications
We are investigating sensing technologies for vehicles through high-speed, high-accuracy recognition of vehicles and their surrounding environment using high-speed vision. For example, we are developing systems such as dynamic vehicle posture estimation and localization and highly accurate inter-vehicle distance/velocity/acceleration estimation for the next-generation’s intelligent transport systems.

・High-speed, dynamic manipulation
We focus on flexible object manipulation, which is considered to be difficult for robots to perform, and we aim to achieve dynamic and high-speed manipulation of flexible objects. In previous research, we achieved one-handed knotting of a flexible rope and dynamic folding of a cloth using a high-speed robot hand system. In addition, we achieved high-speed, dynamic manipulation (for example, ring catching) by a robot arm.

・Human-robot interaction
We are conducting research on human-robot cooperation, assistance systems and human motion enhancement using a super-low-latency and real-time human-robot interaction system implementing high-speed vision and high speed-robot hands. As an example of such implementation, we have developed a janken (rock-paper-scissors) robot with a 100% winning rate.

Text: Yuji Yamakawa (Associate professor)
English proofreading: David Buist (Senior project specialist)