Pictures and descriptions of our lab equipment


Robots
Robots
Hand
Hand
Phantoms
Phantoms
VR
VR
VR
Mobile Robots
Image proc  HW
Image proc.

Robot arms

In the UR vision and robotics lab we have two 6 axis PUMA 760 series robot arms, Bill and Hillary. They are larger versions of the perhaps more popular 560 series arms. The joints are driven by geared electric motors. Each motor is controlled by a digital controller. The robots come with a rather bulky control rack containing the power supplies servo amps, controllers, and an lsi11 central processor. Unimation, the manufacturer delivers the VAL robot programming language which runs on the lsi11. In most of our work we don't use VAL. Instead we use RCCL, which we have ported to run under Solaris threads on an 8 processor Sun Sparcrcenter 2000. Integration with the other vision lab computers is provided through a distributed robotics library implemented in PVM, and high level acess is availible through MATLAB interfaces to the underlying C-code.
Hillary (left), the smaller of the two is a PUMA 761. The elbow link is slightly shorter than Bill's. Bill (right) is the first of our PUMA's. It is a 762 with the extended (850mm) elbow. Click on the images to see larger robot pictures.

Local libraries:


Robot hand

We have a four finger Utah/MIT hand. The hand has 16 actuated joints. Each is pneumatically driven by two antagonistic tendons. The tendons go through a long remotizer, which allows the hand to rotate approximately like a human wrist, but also makes precise control of the fingers difficult.
The hand is mounted on our PUMA 762, with the remotizer bolted to the elbow link. This retains 5 of the 6 freedoms of the PUMA arm. Only joint 6 rotations are significantly restricted.

Force feedback devices: Phantoms

Haptics-enabled virtual reality (VR) applications are much more compelling than graphics-only VR for many tasks (e.g. games, picking and moving molecules and driving simulators). Haptic devices give you the sense of touch through applying forces on certain parts of your body. Traditional haptic devices ( like exoskeletons ) are bulky, expensive and inconvenient to use or program. Originally developed at MIT, Sensable Technology's PHANToM interface is a breakthrough in this field and has gained wide acceptance in research communities.
PHANToM is a desktop device.Basically, the user places his fingertip in a thimble. In each cycle first the fingertip's 3D position is sensed and then a 3D force vector is computed and applied to the fingertip. This whole cycle is repeated at a frequency of about 1000 Hz to give user a smooth (no vibrations felt) force feedback when he moves his finger across space. For more information , the following links can give you a tour of the features of and the many of the exciting on-going and future researches enabled by PHANToM.

Our virutal reality lab currently has four PHANToMs connected to SGIs for studying how human visuo-motor systems work under controlled virtual environments. Another promising research direction is to explore the use of PHANToM as a programmable force-feedback 3D input device such as a 3D accelerator for navigating in virtual environments.


Virtual Reality

VR Lab Hardware Components
  • An SGI Crimson processor, with VGX graphics hardware, for graphics and virtual reality generation.
  • A binocular virtual reality helmet, specially modified to include eye-tracking capabilities inside the helmet.
  • An instrumented ``car'' (on a go-kart frame) with outputs from the steering wheel, accelerator, and brake position.
  • A head-mounted eye-tracking device (including A Flock of Birds sensors for head and hand position). It can track eye movements in head coordinates, and in conjunction with the Flock of Birds head position sensor it can quantitatively track gaze directions in LAB coordinates.
  • High-quality 8mm editing tape deck for recording eye-movement data.
  • A Silicon Graphics Reality Engine for scene generation in the simulated world;

Mobile Wheelchair Robots

We have two wheelchair mobile robots purchased from KIPR, which have been enhanced with extra sensing and computational power by our mobile robotics group. Description of UR mobile robot hardware and software

Image processing hardware

We have two generations of Datacube image processing hardware. An older MV10 board system is on the way out (the boards biting the dust one after another), and a newer mv 200 is being phased in. Either system is capable of a variety of operations at frame rate, such as convolutions, multiplexing, logical operations. Two advantages of the newer systems are: (1) It has an electronic video bus switch (instead of the cable mess seen to the right in the image). (2) One can write "pipelined" code using Imageflow. A Datacube MV200 and ImageFlow User's Guide and several example programs are availible.

Back to URCS Vision and Robotics Home Page


Martin Jägersand                 Comments?