CHANNELS

Aerospace

Automotive

Automated & Autonomous Vehicles

Battery & Electrification Technology

Defense

Drone TV

Electronics

Manufacturing & Prototyping

Materials

Medical

Motion Control

RF & Microwave Technology

Robotics & Automation

Sensors & IoT

Test & Measurement

Aerospace
Automotive
Automated & Autonomous Vehicles
Battery & Electrification Technology
Defense
Drone TV
Electronics
Manufacturing & Prototyping
Materials
Medical
Motion Control
RF & Microwave Technology
Robotics & Automation
Sensors & IoT
Test & Measurement
Aerospace & Defense
Search Results

Vulcan: Intelligent Robo-Wheelchair Navigates Autonomously

University of Michigan computer science and engineering professor Ben Kuipers created Vulcan a few years ago, and now graduate student Colin Johnson and robotics researcher Jong Jin Park have added innovative new capabilities. These capabilities include: the ability to learn the spatial structure of the environment, which allows the robot to efficiently represent its opportunities for travel in the environment; the ability to plan and carry out motion in the environment, avoiding both static and dynamic obstacles; and using its perceptual capabilities to handle the kinds of perceptual problems that arise in natural human built environments. Vulcan can identify 4,800 changing pathway scenarios per second and uses two LIDAR sensors, giving it a 360° field of view of a rapidly changing environment.

Topics:
Automation Electronics & Computers Lasers & Laser Systems Medical Motion Control Photonics Robotics Sensors

More From SAE Media Group

    Transcript

    00:00:05 [Music] we've built a robot that can explore an environment and build its own cognitive map of that environment which it can then use to get around from place to place we've implemented this robot as an intelligent robotic wheelchair we've modeled its spatial knowledge after the spatial knowledge of a human being and we use what we call a spatial semantic

    00:00:33 hierarchy representation that includes both geometrical and topological knowledge our goal is to learn about people's environments and then navigate around them in what we call Campus scale environments the environments are both indoor and outdoor where there are multiple buildings they're linked by paths sidewalks roads one of the characteristic of these environments is

    00:01:03 that you can't count on GPS to tell you where you are so the robot has to be learning the map and keeping itself localized and oriented by using its own perceptual abilities so that's been a major focus of our work my involvement with the wheelchair project is both on the engineering side as well as the reearch side so from an engineering perspective we need to get this 300 lb

    00:01:28 wheelchair robot moving around the environment safely even though people are moving around it so we have to establish you know careful communication and timing between the system and the sensors and the planning modules and the robot needs to know where it is the entire time in order to actually move safely and get where it needs to go my uh work is theoretically uh based on uh

    00:01:51 what's called Model predictive control and it's actually following a principle that every human driver should follow so it identif I how the environment looks like uh where potential hazards are and it tries to predict what's going to happen in the near future maybe for a few seconds and it decides given current situation what is the best action to take at the moment and execute that

    00:02:15 action and repeat and it replans those sequence five times a second at every second it evaluates on average 4,000 possible trajectories and select the best one we have provided two distinct ways for directing the robot in one the human specifies a destination says take me to the front door of the building we assume that the robot has enough knowledge of

    00:02:42 the building to know where the front door is and to know how the structure of the building is connected so that it can get there there's another way that a person might feel comfortable commanding the robot where the instructions are of the form go to the next decision point point and turn right the human is expected to understand the structure the robot has the responsibility and the

    00:03:09 authority to travel along path segments to get from one decision point to the next and and is then told how to make that decision point at the next [Music] place if you really want to change someone's life in in a positive way the easiest thing you can do is give them Mobility where they couldn't before you know we we take advantage of being able

    00:03:33 to walk we don't even think about it we get up we go get a drink we go to the bathroom we go get a cup of coffee but for some people that takes all of their thought process everything they can do is just to move around so if the wheelchair can assist that then they can one gain their independence but two get a much richer experience of life doing robotics in general it's just exciting

    00:03:55 you can actually see the result of your research right away so you write some code you run it your robots either going to hit the wall or avoid people nicely uh it all depends on your theory uh and how you implemented that I have always been interested in how the mind works knowledge of space objects actions that's a kind of Common Sense knowledge by trying to understand how that

    00:04:24 knowledge is structured and how it can be learned by the agent's own experience has always struck me as being one of the great scientific problems [Music]