Mars Mission OnBoard Planner and Scheduler

1
Mars Mission On-Board Planner and Scheduler

• Overview
• The MMOPS Team Ruth Aylett, Les Baldwin, Derek
  Long, Roger Ward, Graeme Wilson, Mark Woods
• Technical Officers Raffaele Vituli and David
  Jameux

2
MMOPS

• 2.5 year study
• Carried out by
• SciSys
• University of Strathclyde
• Heriot-Watt University
• Investigated the suitability of using
• Planning and Scheduling Technology for -
• Autonomous, on-board timeline management in
  robotic Mars Missions

3
MMOPS
Mars Mission On-Board Planner and Scheduler
Introducing On-Board Autonomy for Robotic
Exploration Missions using Planning and
Scheduling Technology
Final Presentation
4
Consequences

• Difficult Mission Planning Environment
• Robust timelines Difficult to Create
• Can lead to Conservative Approach to Planning
• Under utilisation of Resources
• Limited flexibility
• Soft Failures often lead to Suspension of
  Science Activity
• In-efficient Payload Utilisation
• Bottom Line Less Science Carried Out
• Post MER the advantage of mobility has been
  underlined
• Hugely successful but note average travel
  10m/sol. (Sojourner 1m/sol)
• The next set of missions have challenging science
  schedules which will be difficult to meet using
  current technology
• Objective to Maximise Science Return

5
Autonomy

• On-board Autonomy proposed as a solution
• Candidate functions include
• Navigation
• Timeline Management
• Target Selection
• Instrument Placement

6
Autonomous Timeline Management
7
Assumptions
Instruments
Power
Thermal
Memory
8
Dealing with Conflict
Comm.s
NAV
? timeline
Camera
ARM
XRS
t
Instruments
Power
Thermal
Memory
9
Autonomous Timeline Management
10
Background

• Third study looking at application of AI Planning
  and Scheduling to the Aurora Programme

Planning and Scheduling for Aurora Roadmap
11
Unfinished Business - Net Benefit ?

• Deep space, robotic missions have implicit
  autonomy drivers, e.g. communications
  constraints, environment uncertainty etc.
• AI Planning and Scheduling Technology has the
  potential to implement this autonomy
• Earlier studies had demonstrated technology
  potential
• Key question
• Could it add net benefit to a Mars mission such
  as the ExoMars Rover by evaluating it in a
  suitably representative environment.

12
Objectives

• Determine what level of practical on-board
  autonomy is required for an ExoMars like mission
  autonomy philosophy
• Build a prototype application and evaluation
  facility and fly it in a representative
  platform and operations environment
• Demonstrate compatibility with MUROCO-II framework

13
Approach
Operations Experience
Lander Software
B2 Software Test Facility
SCOS 2K MCS
B2 Mission Planning
14
Evaluation Methodology
Compare Trial Results
SCOS MCS
SCOS MCS with Additional MMOPS Components
Beagle 2 Platform and Payload Software Test
Facility
Beagle 2 Platform and Payload Software Test
Facility
Planning And Scheduling
Beagle 2 Lander Software
Beagle 2 Lander Software
Execute Scenarios
15
Autonomy Level Definition

• What level is appropriate for a near-term Mars
  mission ?
• First of all necessary to frame it in terms of
  current ECSS definition of autonomy

16
Space Segment Autonomy

• ECSS E-70-11 defines space segment autonomy
• On-board autonomy addresses all aspects of
  autonomous functions that provide the space
  segment with the ability to continue mission
  operations and to survive crises without relying
  on ground action.
• Autonomy levels for execution of mission
  operations
• execution of pre-planned mission operations
  on-board - traditional
• execution of adaptive mission operations on-board
• Extent of ESA missions to date makes use of PUS
  Event/Action and Scheduling Services Current
  Services have good features but limitations for
  ExoMars context
• execution of goal-oriented mission operations
  on-board
• Foreseen as necessary but considered too big a
  step for near-term missions
• Autonomy levels for fault management
• autonomy to safeguard the space segment or its
  sub-functions
• autonomy to continue mission operations.
• Current suite of PUS services essential but not
  designed to allow global reconfiguration of the
  timeline in response to failure conditions

17
ECSS Autonomy Levels Current
Pre-Planned
Pre-Planned
t
Adaptive
t
18
Limitations and Observations

• PUS Services are the backbone of FDIR and
  operations autonomy
• There are limitations

19
ECSS Autonomy Levels - Predicted
Pre-Planned
Pre-Planned
t
Adaptive
t
Goal Orientated
Goals
t
20
Proposed Service Functionality

• Timeline Validation
• Validates uplinked Timeline against resource
  constraints
• Detects problems with resource changes not
  foreseeable by ground
• Timeline Control
• Monitors executing Timeline against projected
  resource constraints
• Foresees impending problems caused by resource
  depletion
• Responds to actual problems caused by equipment
  failure
• Timeline Repair
• Removes non-viable activities and allow timeline
  to continue executing in planned order
• Inserts additional pre-defined activities to use
  spare resources (either through failure of
  removed activities or beneficial power usage)
• Re-orders activities within constraints and
  dependencies

21
Services Visualised
Validation
Repair
TVCR
Reserve/Opportunities
Priorities, Relationships Constraints
22
Context On-Board
OBS
Device Drivers, HW Interfaces
Rover Hardware
23
Context - Operations
OBS
Distributed Operations
Constraints Input and Management
Ground Segment
SCOS and Mission Planning Applications
24
Evaluation Architecture
(TSIM)
(TSIM)
IO Module
IO Module
MUROCO
-
II
MUROCO
-
II
Platform and Payload
Platform and Payload
MMIM
ARE
MMIM
ARE
Simulation (PPS)
Simulation (PPS)
SUN Solaris
SUN Solaris
TEST with TVCR
TEST with TVCR
TEST without TVCR
TEST without TVCR
i.e. Nominal Beagle 2
i.e. Nominal Beagle 2
System
System
PC/Laptop/Linux
PC/Laptop/Linux
SCOS 2000
SCOS 2000
ECCS
ECCS
TC
TC
TM
TM
TM Displays
Manual Stack
TM Displays
Manual Stack
Scheduler
CONTOOL
Control Manager
Scheduler
CONTOOL
Control Manager
Addition of Constraints
Creation of Initial Timeline
Addition of Constraints
Creation of Initial Timeline
Structure
Structure
Structure
And opportunity science fragments
25
Test Trials

• Using detailed (aprox. 30 Activity
  Sequences/OBCPs) two sol Beagle 2 scenario as
  nominal test timeline
• Scenario based on a rock analysis and involves
  recurrent use of
• ARM
• Mossbauer
• X-Ray Spectrometer
• Camera
• Rock Corer Grinder
• Microscope
• Various fault scenarios injected to evaluate TCVR
  capability

26
Results
Opportunity Fragment Inserted
Fragments Removed
Fault Detection
27
MUROCO-II

• Objective to demonstrate basic interfacing
  between TVCR at timeline level and MUROCO-II at
  task level
• Developed a simple interface between TVCR and
  MUROCO-II ARE Simulator
• Successfully demonstrated using TVCR to repair a
  timeline on-board the ARE containing MUROCO-II
  like tasks

28
General Conclusions

• AI Planning and Scheduling can be used to
  implement autonomous, on-board timeline
  management
• Addition of TVCR service will increase payload
  utilisation increase net science returned
• TVCR Function demonstrated in a highly
  representative environment
• Migration to Flight representative hardware not
  trivial but possible as a Phase B1 type activity
• Approach appears to be consistent with a graceful
  evolution of existing Mission operations approach
• Successfully demonstrated using TVCR to repair a
  timeline on-board the ARE containing MUROCO-II
  like tasks

29
Observations

• Requires a system level view/involvement from
  early development phases
• Closer integration with existing state assessment
  services should be considered
• Development of space Mission planning support
  tools i.e. final version of CONTOOL and TVCR
  should not be divorced

30
Future Possibilities for Autonomy

• CREST Activity
• Development of a closed-loop science capability
  for a rover
• Autonomous detection of the target
• Assessment of response
• Decision
• Implementation e.g. instrument placement

31
Dust Devil Detection