Human Role in Lunar Landing

1
Human Role in Lunar Landing

• Charles M. Oman, Ph.D.
• Director, Man Vehicle Laboratory
• Massachusetts Institute of Technology
• Sensorimotor Adaptation Research Team Leader
• National Space Biomedical Research Institute
• coman_at_mit.edu
• Go for Lunar Landing
• From Terminal Descent to Touchdown
• Tempe, Arizona, March 4 2008

2
Human Role in Lunar Landing

• Technology has improved since Apollo but human
  brain has not.
• What is proper allocation of tasks between human
  and machine ?
• Apollo 11 workload was 13 on a scale of 10.
• Crew needs support for autonomous time critical
  decision making, e.g. landing site
  re-designation, abort, failure diagnosis.
• Should pilot have final control authority, or
  just a vote ?
• Will pilots trust automation ?
• Automation surprise Why did it do that ?...
• Are modes intuitive ? Whats it doing now ?
  What next ?
• Is automation clumsy to re-program ?
• Can crew gracefully revert to lower automation
  levels, or must they revert to full manual?

3

• Fully automatic landings to within 10 m of a
  touchdown point at well surveyed sites will be
  technically possible.
• Safety and certification costs considerations
  dictate capability for manual control of flight
  path and attitude (NASA HRR 8705.2A34495)
• Crew will visually evaluate touchdown area, then
  approve, re-designate, or fly manually.
• Manual flight will likely remain the operational
  baseline.
• As on Shuttle Train as you fly fly as you
  train
• Astronauts are pilots and explorers, not cargo.
• Crew will have HUDs, cockpit map, profile,
  perspective terrain displays of 3D DEM terrain
  data and trajectory to enhance situation
  awareness and prevent geographic disorientation.
• Flight path predictors, recommended touchdown
  zones, and fuel range circles will aid decision
  making.

4

• 0.5 m terrain features arent visible to the
  naked eye till 4000 feet away. (5 arc min vs .5
  for lidar)
• Humans have difficulty judging surface slope,
  smoothness, shape, and size
• Shape-from-shading Crater-Illusion craters
  seem convex, rather than concave when viewed
  looking down-sun.
• Shadow size depends on sun elevation.
• Surface slope may be misjudged due to
  non-Lambertian regolith reflectance, weak
  gravitational down cues.
• Crater, boulder, and rock size is ambiguous
  they occur in all sizes.
• Harder to judge distance due to lack of
  atmospheric light scattering. Distant objects
  have unnaturally higher contrast, darker color.
• Does Earthshine alone provide sufficient
  illumination ?
• Sensor images may be difficult to interpret.

5

• Handling qualities will be superior to Apollo LM.
• Translations require large vehicle rotations.
  When visual or vestibular cues are ambiguous,
  sensory PIOs are possible.
• During final descent
• Crew cannot see terrain directly beneath. Apollo
  style slow forward translation keeps touchdown
  point in view.
• Dust grayout below 50-100 reduces visual
  altitude, attitude and translation cues (e.g.
  A12, 15 helicopter brownout accidents)
• Streaming dust can create visual illusion of
  backward movement. LLTV did not model lunar
  visual environment.
• Conclusions
• Early human-in-the-loop simulation is critical
  for automation development. Reduces subsequent
  need to train around design deficiencies !
• Simulators must accurately model both visual and
  motion cues.
• Research needed on e.g. sim dust models, motion
  washout, sensory PIOs.

6
(No Transcript)
7
Backups
8
Distance and Slope Judgment
9
Shape from Shading

• Which crater appears concave ?
• Shape is inferred from shading employing a light
  from above assumption - even in orbital flight.

(Oman, 2003)
10
Geographic Disorientation

• Apollo 15

• A15 crew realized they werent heading for
  planned spot, and didnt know exactly where they
  were relative to any familiar landmarks. So
  they picked a smooth area nearby and headed for
  it. (Mindell, 2007)
• The problem was, when we pitched over and began
  to look out the window, there was nothing there
  !
• I was very surprised that the general terrain
  was as smooth and flat as it was..there were very
  few craters that had any shadow at all, and very
  little definition. (Dave Scott)

11
Landing Zone Assessment

• Apollo goal touchdown on lt 5 deg slope, lt 2 ft.
  variations
• Perceptual limitations
• Cognitive map includes only large landmarks.
• Fractal terrain, difficult to remember/recognize.
• 0.5 m landmarks become visible at 4000 feet.
• Regolith reflectance is not Earthlike
  (non-Lambertian)
• Slope difficult to judge at steep visual angles
• Shading elevation cues are ambiguous.
• Light from behind/below can make craters appear
  convex

Apollo 15
12
Dust Grayout

• Grayout at lt 50-100 causes progressive loss of
  horizon, altitude, position cues.
• I couldnt tell what was underneath me I knew
  it was a generally good area and I was just going
  to have to bite the bullet and land, because I
  couldnt tell whether there was a crater down
  there or not.
• It turned out there were more craters there than
  we realized, either because we didnt look before
  the dust started or because the dust obscured
  them

Pete Conrad, Apollo 12
13
(No Transcript)
14
Touchdown Terrain Awareness

• Crew cannot see below and behind. Must remain
  aware of the terrain beneath during descent.
• Apollo 15 landing gear overlapped edge of a small
  crater. Descent engine bell damaged by crater rim.

15
STS-3 unexplained PIO
16
Manual vs. Automatic

• Constellation Lunar Lander will have autonomous
  landing capability - needed for uncrewed
  operations.
• NASA Spacecraft Human Rating Requirements require
  capability for manual control of flight path and
  attitude.
• Apollo 14-17 LEMs had autoland capability -
  though it was never used. Why ?
• John Young Because the place we were landing
  was saturated in craters and the automatic system
  didnt know where the heck the craters were, and
  I could look out the window and see them. Why
  trust the automation anyways? Youre responsible
  for the landing. You know where you want to land
  when you look out the window and why dont you
  make sure you land there?

(Cummings, et al 2005)
17
Semi-Autonomous Option

• ALHAT method (in development)
• Lidar scans terrain at 7K, automation suggests
  landing spot

too far for terrain sensors
too high for window view
too far for human eye
7000
4000

• Crew visually confirms site and either
• approves
• redesignates, or
• flies manually

too shallow for terrain sensors
(Forest et al, 2007 Brady, et al 2007)