Sensorbased Human Motion Tracking Drew Walton CS 525
Sensor-based Human Motion Tracking Drew Walton CS 525 Project Dr. Chow UCCS
Overview Describe the reason for evaluating different types of monitoring technologies Define requirements for consideration as a good choice for monitoring body movement Introduce four technologies that are capable of performing the required functions Compare the selected technologies
Reason for this project Interesting problem Curiosity Remote monitoring of rehabilitative exercises! What technologies are available? How are they being used for this application? Is there a problem to solve? Or am I too late?
The Present Situation Numerous technologies exist to monitor the movement of the human body Mechanical Sonic Magnetic Accelerometers and gyros Video Combinations
Requirements: minimum Track all types of motion Track through realistic velocities of motion Accomplished by adequate sample rates Track through all ranges of motion Bending, twisting, circular, jumping No motions should be out-of-bounds Tracking must be accurate Accuracy must be fine enough to capture subtle movements
Requirements: Nice to have Non-bulky Easy to wear Not all patients will be technical Reasonable price Not all patients will ave same range of motion, strength or dexterity Easy to use Must not interfere with the exercise Doctors will want to own many Don't want them to “disappear” Rugged Software to display the movement
Technology #1: Crossbow “Mote” Uses a combination of sonic and radio frequency transmissions Similar to counting the time between lightning and thunder Somewhat large circuit boards act as transmitters or receivers Currently best at tracking an entire person Multiple transmitter, single receiver approach
Technology #1: Crossbow “Mote” Limitations Not very accurate Suffers from multi-path issues Too large to wear comfortably Not able to make rapid samples Can't use at all on small body parts Too slow to track quicker movements No software available to display the body moving One hand-held, or serial port needed for each receiver
Technology #2: MIT Body Suit Uses ultrasonic transmissions, accelerometers and gyros Very accurate Uses 8 transmitters, 18 receivers all mounted on a body suit Must carry all equipment in a backpack Display software is available from somewhere
Technology #2: MIT Body Suit Limitations Heavy backpack Can't target single area to track Suit may be difficult for some to put on There is potential for accumulated errors to result in inaccurate displays
Technology #3: Moven Body Suit Very accurate Uses only inertial sensors, immune to noise Wireless transmission of data to base PC Body suit can be worn whole, or select upper or lower body Comes with display software
Technology #3: Moven Body Suit Limitations Still is a suit: some people may not be able to put on Cost is very high: $50, 000
Technology #4: Vicon video The most accurate Uses 8 special video cameras all around the person being tracked No suit required! Patient wears small sensor dots for camera to pick up Display software is very nice Even has a language equivalent to Labanotation
Technology #4: Vicon video Limitations A lot of equipment Price! As much as $194, 000 Cameras and mounting gear Receivers Computer A lot for a patient to set up
Comparison Accuracy: All have but crossbow, and MIT has cumulative error Bulkiness: Vicon is best, but Moven suit is very good Easy to wear: Vicon is dots placed using glue or velcro. Moven suit looks very good though Easy to use: Moven seems easiest. Cost: Crossbow is cheapest, but it doesn't really work. MIT is next, but isn't a product yet. Software: Vicon is best, Moven is good. MIT is from unknown source
References http: //www. xbow. com (Crossbow Industries) http: //www. thinkartificial. org/artificialintelligence/whole-body-computer-interfaces/ (MIT Motion capture suit) http: //www. moven. com (Moven motion capture suit) http: //www. vicon. com (Vicon video motion capture system) http: //www. motion-capturesystem. com/resources/hardware. html (Listing of many companies and their motion capturing technologies)
- Slides: 16