Robotics Institute
Seminar, October 8
Time
and Place | Seminar Abstract | Speaker
Biography | Speaker Appointments
The Mars Exploration Rover Descent Image
Motion Estimation System
Dr. Andrew E.
Johnson
Senior Member of Technical Staff
Jet Propulsion
Laboratory
Mauldin Auditorium (NSH 1305)
Refreshments 3:15 pm
Talk 3:30 pm
Nineteen months
before the launch of the Mars Exploration Rovers (MER), it was discovered that
sustained horizontal winds during descent could cause the lander to impact with
enough force to tear the airbags and damage the rover payload. At this point,
the MER Entry Descent and Landing (EDL) system had small transverse rockets
that could point the large retro rockets to reduce the horizontal speed of the
lander, but it did not have a sensor for measuring horizontal velocity. Because
incorporating a traditional radar-based velocity sensor into the relatively
mature EDL mechanical and electrical design was infeasible, the MER Project
decided, against the advice of many, to pursue an unprecedented camera based
approach to velocity estimation. Thus began the accelerated development of the
Descent Image Motion Estimation System (DIMES), the first autonomous machine vision
system used during planetary descent.
DIMES
is composed of sensors and software including a descent imager, a radar
altimeter, an inertial measurement unit and an algorithm to estimate horizontal
velocity. This algorithm combines radar,
image and inertial data in novel way to create a low cost, robust and
computationally efficient solution to the horizontal velocity estimation
problem. This talk will start with an overview of the MER mission and then
describe the DIMES hardware and software development, the DIMES field tests
over Mars analogs in the Mojave Desert and
the successful performance of DIMES during the Spirit and Opportunity
landings in January 2004.
Dr. Andrew
E. Johnson graduated with Highest Distinction from the University of Kansas in 1991 with a BS in Engineering Physics and a BS in Mathematics. In
1997, he received his Ph.D. from the Robotics Institute at Carnegie Mellon University where he developed the spin-image surface signature
for object recognition and surface matching. Currently, he is a Senior Member
of Technical Staff at NASA’s Jet Propulsion Laboratory where he is
developing image-based techniques for autonomous navigation and mapping during
descent to planets moons, comets and asteroids. At JPL, Dr. Johnson has worked
on technology development tasks as well as flight projects. For the Mars
Exploration Rover Project, Dr. Johnson was the lead algorithm developer for the
Descent Image Motion Estimation Subsystem (DIMES), the first autonomous machine
vision system used during planetary landing. Following the successful
development and execution of DIMES, he is now moving back to the development of
machine vision systems for landing hazard avoidance, pin-point landing and rover
navigation. Part of this work includes a collaboration with the University of Southern California and the University of Minnesota in the area of vision guided safe and precise landing for autonomous
helicopters. In 2003, Dr. Johnson was awarded the JPL Lew Allen Award for
Excellence for “his groundbreaking contributions in the area of machine
vision algorithms for safe and precise landing.”
For appointments, please contact Martial Hebert.
The Robotics Institute is part of the School of Computer Science, Carnegie Mellon University.
