Main Page | Recent changes | Edit this page | Page history

Printable version | Disclaimers | Privacy policy

Not logged in
Log in | Help
 

Spring W10 Field Test

From DGC Wiki

This page describes the goals and objectives for the Spring W10 Field Test, scheduled to take place on 5-4 June at Stoddard Wells. Instructions for editing this page:


</table>

GOTChA Blockers Schedule Field Notes Spring Test Plan

</tr> </table>


GOTChA Chart

Goals

  • P Demonstrate planning based on integrated road following [Did several miles of following road trajectories; not map-based planning]
  • ✔ Demonstrate planning based on integrated static data (GPS traces) [Proof-of-concept demonstrated for short sections]
  • P Demonstrate complete system architecture (including deliberative/reactive (separately), modeman, GPSFinder, roadFinder, ladar/stereo) with occasional obstacles [Tested deliberative and reactive with ladar in Gazebo; stereovision was not able to be calibrated; state and map issues prevented significant integrated testing]
  • X Enable offline testing and module characterization with logged and simulated data playback in modules [not yet ready]


Prep

Day 1

Day 2 </table>

Test Team

  • Pfister (IPT)
  • Foote (veh/emb, bld)
  • Hines (veh/emb, tst)
  • Pollack (veh/emb, ops)
  • Stewart (pln, doc)
  • Craig (pln, doc)
  • Yosinski (pln, test)
  • Soudek (ter, mod)
  • Heyneman (ter, mod)
  • Feingold (ter)
  • Sexson (veh/emb)

</table>

Objectives

Critical:
  • ✔ E-stop fully operational & software failsafe (Tully)
  • X Centralized and unified capability to log all sensor data (hardware limited) (Jason)
  • X Capability to play back logged and simulated data with repeatable results (Jason)
  • X Deliberative planner perfoming obstacle avoidance at 20 mph (Jason)
  • ModeMan able to slow trajectory to zero and keep it there (Alex)
  • ? Reactive Planner able to obey same continuity constraints as deliberative planner (Alex)
  • X Planning with static data feeding seeds (Dan)
  • ✔ Road following operational (unit test & integration) (Dan)
Baseline:
  • X Ladars, Riegl operational (interfaced) (George)
  • Installation & demo of new steering motor (Ben)
  • Installation of steering motor gearbox & new steering motor driver interface (Ben)
  • Transmission/transfer case & ignition actuation (Harmon)
  • Demo of gimbal platform to evaluate gimbal performance
  • ✔ Static data required for all of intended testing area (Dan)
  • Mount, interface and test Riegl and second Sick Ladar (George)
  • High-speed driving (>20 mph)
  • X Minimize AState IMU drift (able to run without GPS with some TBD performance spec) (Tully)
Bonus
  • X 50 miles of autonomous driving with complete architecture and occasional obstacles
  • X Full course completed without stops in Gazebo

</table>

Technical Challenges (blockers)

  • Strike out when resolved

AState

  • Characterize angular drift (W9)
  • need to be able to run X>0 time w/o GPS with
    • less than Y rad/sec drift
    • less than Z mm/sec drift
  • Raw data playback on time cue from logger module

ADrive

  • transmission interface
  • new steering motor interface

Road Finding

  • get code working in our repository (Bug #1445)
  • output angle to vanishing point (W9)
  • filling in fusion mapper (Bug #803)

TrajFollower

  • Fix feedforward constants (W8) (bug #1258)
  • Lateral error < 30cm on dirt at 20mph

Planner

  • doesn't slow to a halt fast enough
  • clean up obstacle grower

ModeMan

  • fix traj checking
  • fix trajectory slowdown (W9)
  • test with reactive and planner together, with one traj evaluation function, always sending the best traj to follower

Reactive Planner

  • Test on Alice during W9 with planner/modeman

GPS Finding / Static Mapping

  • GPS finding providing a seed for the planner

Fusion Mapper

  • implement non-binary obstacles (bug #1707, bug #1706)
  • fuse multiple data sources into elevation layer
  • fuse multiple cost layers (road, terrain, RDDF, GPS) into combined cost
  • test and benchmark alternative map resolutions
  • remove obstacle grower
  • Add option to mapper to grow obstacles in the combined cost layer (bug #1262)

Data Logging

  • raw data and skynet message logging
  • playback of skynet messages
  • raw data playback in appropriate modules

Gazebo

  • HOWTO document for Gazebo
  • solve simulated state issues
  • Riegl mounted

Ladar

  • increase speed from 7Hz to 75Hz (bug #793, bug #1639)
  • Raw data playback on time cue from logger module (or from state playback?)

Vehicle

Embedded

  • validate guaranteed packet functionality (bug #1363)

Ops

  • get Harvey ready to go
  • wireless bridge for chase vehicle

Bugzilla Links

Bugzilla blockers for Spring W9+W10 field test]
All Bugzilla tasks for Spring W9+W10 field test]
Bugzilla blockers for Spring W9 field test]
Bugzilla blockers for Spring W10 field test]

Approach (timeline)

Mo 5/16 5:00p Test team meeting to agree on goals and objectives
Tu 5/19 12:00p Test team meeting Finalize goals and objectives
Tu 5/24 12:00p Review test goals at project meeting
Tu 5/24 1:00p Test team meeting - Update blockers / bugzilla for W10, Finalize plan for W9


Tu 5/31 1:00p Test team meeting - check on status
Th 6/02 5:00p Test status meeting - go over test status; determine actions + time of go/no-go meeting
Sa 6/04 7:00a Go/no-go meeting - verify ready to go
Sa 6/04 Test day #1
  • Setup base camp, operations
  • Demonstrate critical and baseline objectives
  • End of day meeting @ 7pm (in field)
Su 6/05 Test day #2
  • Demonstrate baseline and bonus objectives
  • Return to Pasadena by ~8 pm
Su 6/05 8:00p Test review (all test participants; restaurant)

</table>

</table>

Test Schedule

Prior to Go/No-Go Meeting

For checkout purposes, last stable build: <insert revision # here> (Tully)
  • Equipment preparation (TBD)
    • List any unusual items that need to be taken
  • Simulation of all tests on Gazebo (TBD)

Top
GOTChA
Prep
Day 1
Day 2
</table> </table>

Specific tests to run before leaving shop

  • TBD

Day 1

  • Test Attendance Saturday:
    • Veh/Emb : Sam, Jeff, Tully
    • Planning : Lars, Chris W., Adam C., Jason
    • Terrain : Ryan F, Dan, Jeremy L, Jeremy G
Goals
  • Establish initial operations
  • Complete critical objectives
Test Manager: Pfister
  • Run all meetings (go/no go)
  • Maintain (and realign) schedule
  • Team member marked in bold is the task leader
  • Team member marked in italics is the default notetaker

</table>

Top
GOTChA
Prep
Day 1
Day 2
</table> </table>

TimeTrack 1Track 2Operations
7:00a
Go/No-Go Meeting
  • Go through pre-conditions and make sure we are ready to leave
 
  • Transfer databases to lawrencium (Owner)
8:00a
Drive to test area
 
  • Sam, Jeff, Tully, Lars, Chris W., Adam C., Jason, Ryan F, Dan, Jeremy L, Jeremy G
  • Alice (4): Driver, passengers
  • Tow vehicle (2): Driver, passengers
  • Harvey (6): Driver, passengers
  • Additional cars:
  • Jeff's truck (2) Jeff, passengers
  • Tully's car (3) Tully, passengers
  • Directions to [[<Name> Test Area]]

</table>



10:00a
 

A1: astate, adrive (Tully)

  • Use w10_testing branch
  • Speed : as needed
  • Distance : as needed
  • Modules : astate adrive trajfollower GUI
  • Team : Tully, Jeremy L, Jeremy G., Dan
  • Results :Notes A1
    • Unknown problem with Astate or SDS
    • Successful verification of adrive modifications
 

B1: Road Follower #1 (Dan) * Pushed to 11:30. Issue with Dan and Ryan building with old version of Chris R's code

  • Test road following on straight/slight-curved roads
  • Target of about 1 mile of straight/slight-curved roads in various lighting conditions
  • Viewable real time vanishing point info
  • Logging of road vanishing point and image data
  • Team : Tully, Ryan F, Dan, Christopher R.
  • Results :Notes B1
    • Successful road tracking
    • Astate lost gps link and diverged
 

C1: Set up operations area (owner)

  • Set up RV, tables, networking, power
  • Team : Owner, team members




11:00a
 

Stereo cameras out of calibration from drive up

A2: Fusion mapper test #1 (Jeremy G)

  • Use w10_testing branch with ladar feeder on to evaluate spurious data
  • stereo off to start
  • add stereoFeeder as allowable by road following tests
  • Speed : 10-15mph
  • Distance : about 1 mile
  • Modules : astate adrive trajfollower fusionMapper plannerModule ladarFeeder stereoFeeder GUI
  • Team :

A2: AState diagnostics (Jeremy L.)

  • Roll back Astate to w6stable
  • Results :Notes A2

Lunch

Helpers (2): helper1, helper2


12:00a




1:00p
 

B2: Road Follower #2 (Dan) Push the turn characterization til tomorrow. Instead focus on controlling Alice on straight stretches from road detector

  • Road Follower specific driving

  • Test road following characterization entering and leaving hard turns
  • Characterize detection failure and reaquire
  • Viewable real time vanishing point info
  • Logging of road vanishing point and image data
  • Team :
  • Results :Notes B2
    • .7 miles fully autonomous road following
    • biased to right side of road
    • Stopped by astate issues
 

C2: Scout alternate sites (owner)

  • Only if there's an available vehicle
  • Vehicle :
  • Team :




2:00p
 

A3: Fusion mapper test #3 (Jeremy G.) Out at 4:30 with obstacle grower working

  • Use w10_testing branch with ladar feeder on to evaluate spurious data performance
  • stereo on if A4 results are good and ok with road following tests
  • Speed 5,10,15,20 mph
  • Distance : 10 miles total
  • Modules : astate adrive trajfollower fusionMapper plannerModule ladarFeeder (stereoFeeder) GUI
  • Team :
  • Results :Notes A3
    • Removed obstacle growing code (memory leak)
    • Successful evaluation of non binary map with minimal spurious data. Failed to reach 10 miles due to planner failure (later corrected : map resolution mismatch)




3:00p
 

fusionmapper code not ready for road following interface. Try to get Joe out to help. (UPDATE: didn't happen) B3: Road Follower #3 (Dan)

  • Road following interface test
  • Demonstrate merging of road following data into cost-map
  • Viewable cost-map layer
  • Logging of cost-map and image data
  • Team :






5:00p
 

No time to perform, A3 ran til 7:15 A4: Aggressiveness tuning demonstration in fusionMapper (Jeremy G.)

  • Test online tuning of elevation gradient to speed
  • Test online scaling of RDDF speed
  • Use w10_testing branch
  • Speed : 5, 10, 15mph
  • Distance : 1 mile
  • Modules : astate adrive trajfollower fusionMapper plannerModule ladarFeeder/stereoFeeder GUI
  • Team :





6:30p
End of Day Meeting
  • Go over results of the day; replan as needed
  • Define action items for evening
Departures
  • Vehicles : Tully's car (3), Jeff's truck (2), Tow vehicle (2-3)
  • Jeff, Tully, Adam, Chris W, Jeremy G, Jason, Jeremy L, Jason?
7:00p
Dinner
  • Helpers (2): helper1, helper2
 
  • Shuttle: Driver



Evening
Arrivals
  • Vehicles : Ben's car (3), Tow vehicle (2)
  • George, Ben, Alex, Ryan C., Eric
  • Some miscommunication : Ryan wasn't there
Working time

Ryan not here so instead collect static data, get gas

  • A5: Reactive Planner #1 (Alex)
  • Initial test of reactive planner
  • Gather as much information as possible for overnight/morning debugging
  • Team : Ryan C.

 
  • Possible run into town for supplies, showers, etc

</table>







Day 2

  • Test Attendance Sunday:
    • Veh/Emb : Sam, George, Ben
    • Planning : Lars, Ryan C, Alex, Jason?
    • Terrain : Ryan F, Dan, Eric
Goals
  • Complete baseline and bonus objectives
Test Manager: TBD
  • Run all meetings (go/no go)
  • Maintain (and realign) schedule
  • Supervise filming
  • Team member marked in bold is the task leader
  • Team member marked in italics is the default notetaker

</table>

Top
GOTChA
Prep
Day 1
Day 2
</table> </table>

TimeTrack 1Track 2Operations





8:00a 9:00a
 

Static data not quite ready so do further Astate diagnosis and fusionMapper testing

A6: Astate and non-binary fusion mapper test (Alex)

A6: Static Data #1 (Alex)

  • Test planning based on static data filling in map
  • Use w10_testing branch
  • Speed : 5-20mph
  • Distance : 5 miles
  • Modules : astate adrive trajfollower fusionMapper plannerModule GPSfinding GUI
  • Team : Eric C.


 
 

C3: Scout alternate sites (George) Not enough manpower for actual scouting but reviewed details of sites with Tony

  • Only if there's an available vehicle
  • Vehicle :
  • Team :





10:00a 11:00a
 

Scratch - no Ryan C. A7: Reactive Planner #2 (Alex)

  • Test reactive planner
  • Use w10_testing branch
  • Speed : 5-15 mph
  • Distance : about 1 mile
  • Modules : astate adrive trajfollower fusionMapper reactivePlanner ladarFeeder GUI
  • Team : Ryan C.


 

B4: Road Follower #4 (Dan)

  • Further testing of Cristopher R's code integration
  • Results :Notes B4








1:00p
 

A8: Static Data #2 (Alex)

  • Test planning based on static data filling in map and sensor data
  • Use w10_testing branch
  • Speed : 5-20mph
  • Distance : about 1 mile
  • Modules : astate adrive trajfollower fusionMapper plannerModule ladarFeeder/stereoFeeder GPSfinding GUI
  • Team : Eric C.

 
Lunch

Helpers (2): helper1, helper2





3:00p4:00p
 

A7: Deliberative Planner (Alex)

  • Test deliberative planner
  • Use w10_testing branch
  • Speed : 5-15 mph
  • Distance : about 1 mile
  • Modules : astate adrive trajfollower fusionMapper deliberativePlanner ladarFeeder GUI
  • Team : Alex
  • Results :Notes A7

A9: Planning based on road following (Dan)

  • Test planning based on static data filling in map
  • Use w10_testing branch
  • Speed : 5-10mph
  • Distance : 1 miles
  • Modules : astate adrive trajfollower fusionMapper plannerModule roadfinding GUI
  • Team :

 


5:00p
 

A8: Static Data #1 (Alex)

  • Test planning based on static data filling in map and sensor data
  • Use w10_testing branch
  • Speed : 5-20mph
  • Distance : about 1 mile
  • Modules : astate adrive trajfollower fusionMapper plannerModule ladarFeeder/stereoFeeder GPSfinding GUI
  • Team : Eric C.

A9: Planning based on road following (Dan)

  • Test planning based on static data filling in map
  • Use w10_testing branch
  • Speed : 5-10mph
  • Distance : 1 miles
  • Modules : astate adrive trajfollower fusionMapper plannerModule roadfinding GUI
  • Team :

 

Static Data on Alice

 
7:00p
Pack & Return to Pasadena
8:00p
End of test meeting
  • including dinner
9:00p
Unpack

</table>

Additional Notes


Retrieved from "http://kincora.cds.caltech.edu../../../s/p/r/Spring_W10_Field_Test_1ffb.html"

This page has been accessed 385 times. This page was last modified 19:11, 7 June 2005 by DGC Wiki user Jeremy. Based on work by Sam Pfister and DGC Wiki user(s) Lars, Hines and Klk.


[Main Page]
Main Page
Alice Docs
DGC07
DGC05
Categories
Recent changes
Help

Edit this page
Discuss this page
Page history
What links here
Related changes

Special pages
Bug reports