Robotics


THE CHALLENGE

The proportion of older adults in the world is growing. This global trend of population aging will create a large social and economic burden.

Enabling older adults to age in place longer through technology can reduce the monetary and time cost of elder care.

Grasping and manipulation of household objects

TRI'S APPROACH

TRI develops intelligent, physically capable and socially assistive robots for improving the quality of life of older adults.

Older adults experience higher levels of well-being when they lead an active and engaged lifestyle, even as their health and physical capabilities decline.

The goal is to enable them to lead independent, healthy, socially active and dignified lives.

HOW IT WORKS

Assistive robots are machines composed of subsystems that work together to achieve a task: mechanical mechanisms to move, touch, and hold; planners to reason about the task; mapping and localization to understand the space the robot moves in; sensing and perception to understand the objects in that world; and simulation to predict and understand how the robot and its world might interact.

To create the next generation of assistive robots, TRI is developing technology with enhanced perception, manipulation, learning and reasoning through advances in artificial intelligence and cloud robotics. The technology being developed will be applied to provide assistance in the home.

TRI PROGRESS

Manipulation

TRI is researching and progressing core technology that enables robust manipulation for the home.

Simulation

TRI is advancing simulation capabilities that enable real-world robots to be developed in software. The video on the left shows how TRI's simulation of bimanual manipulation (left side) agrees closely to reality (right side). Additionally, TRI is the primary contributor to Drake, an open source C++ toolbox for analyzing the dynamics of robots and building their control systems with a heavy emphasis on optimization-based design/analysis.

Perception

TRI uses robotic perception, most commonly for the localization of objects and the accurate estimation of their poses (i.e. their rotation and position in the world). TRI has pioneered one of the first works that successfully combines deep learning with synthetic training for "6 Degrees-of-Freedom" object detection. This "SSD-6D" teaches robots to perceive a broad category of objects in real-time, which enables robots to do more complex tasks like dishwashing or playing with Legos.

ACADEMIC PARTNERSHIPS

TRI supports and collaborates with academic institutions and RoboCup@Home teams using Toyota’s Human Support Robot (HSR) as a development platform. We also seek to enable home robotics researchers share ideas and collaborate to develop open source software, working with a number of talented university teams.

WHAT'S AHEAD

TRI continues to advance the state of the art in manipulation, mobile manipulation, simulation, perception and autonomy. TRI leverages these advances to develop products that help older adults age in place longer.