Space Tug Phase B Autonomy

1
Modeling Autonomous Space Tug Rendezvous

• Gergana Bounova
• October 9, 2003

2
Modular Model
Target Information
Target Found (x, y)
Uncertainty Model Representation
Orbital Elements
Setup Grid Local Space Search
Grid Type And Size
Recurrent Orbit Propagation
Coord. Transformation Maneuver Optimization

• Easy model update
• Flexible per module
• optimization

Outputs Fuel Time, Cost
Budgets
Modular Model Rendezvous - Grappling
3
Target Information

• Orbital Elements (or (r,v))
• Generic orbital elements
• - a i ? ? e M
• Target orbital elements
• 7191 45 0 0 0.006 0
• Assumed uncertainty 100 m
• - reflected in true anomaly, eccentricity, and
  inclination disturbances

• Geometry (40 kg)
• diam 1.04 m, hmax 2.23 m

4
Relative Orbits Hills Frame
5
Uncertainty and Representation

• Cylindrical grid derive from uncertainty band
• Set up as 2D grid surface
• Patch extension from previous code
• Fully functional search routine
• Problems? Uncertainty grows as the orbits
  propagate!
• The cylindrical grid does not cover the local
  sky
• If the search is too long, the target vector
  falls outside of the grid

6
Spherical Coordinates

• Earths latitude-longitude analogy
• Grid size function of Field Of View
• Unlimited local sensor range
• Cons
• Singularities, coordinate trans.
• Grid patching case-by-case
• Benefits
• Sky coverage
• Easy sensor positioning, pointing, sensor
  optimization

7
Local Space Search

• Strategy
• Unlimited sensor range allows 2D search
• Previously tested algorithms random sensorless,
  semi-autonomous, autonomous strategies
• Implemented a hybrid approach cyclical
  interchange of random and autonomous steps
• Ideas
• Algorithm optimization simple versus refined
• Sensor optimization
• Outputs to maneuvering
• 2D grid coords (x_found, y_found)
• Target and tug orbital elements (propagated
  during search)

8
Maneuver Optimization

• Linear quadratic relative orbit controller
  RelativeOrbitControl.m (S/C Control Toolbox)
• Toolbox code used works for SMALL perturbations
• Assumes full visibility (GPS, sensor tracking)
• Rarely diverges when the RK4 orbit propagation
  continues for too long
• For the time and space scale used does a great
  job!

9
Controller Convergence
10
Budgets

• Delta V Attitude Rendezvous
• Within 30 m/s
• Time Computation Attitude Rendezvous
  Maneuver
• Usually between 6.5-7 hrs, max 9.5 hrs
• Nearly quadratic relationship between time and dV
• T V2 scaled and translated

11
Future Grappling

• Model grappling physics using CAD functions
• Grappling mechanism design

12
Back up 1 (62x) Description of Hardware

• The Tug uses a Handyboard on-board computer
• - 4 motor ports
• - 16 analog sensors ports
• - Motorola 68HC11 Processor

• Tug sensors
• Long range Infrared
• Short range Infrared
• Touch

• Tug communicates with command computer through
  serial port interface

13
Back up 2 (62x) Test Matrix
Distance to target

• Type of strategy versus Distance to target
• Strategies range
• Random algorithm
• Semi-autonomous with
• human in the loop
• Fully autonomous
• Positions range
• Close target distance within
• 10 of search space
• Half-way within 50
• Far at the other end of the table
• Third Dimension
• Number of runs 5

n
I
II
III
IV
V
VI
VII
VII
IX
Type of strategy
14
Back up 3 (62x) Data Analysis I

• Semi autonomous requires least amount of energy!
• Random search is overall fastest
• Autonomous for space?

15
Back up 4 (62x) A Random Walk