Overview of the Intelligent Vehicles and Systems Group

1
Overview of the Intelligent Vehiclesand Systems
Group
Penn State University by Dr. S. Brennan
See http//controlfreaks.mne.psu.edu for more info
2
An introduction to Sean Brennan

• Youngest faculty with full appointment in ME, 5th
  year currently
• Graduated from the University of Illinois at
  Urbana-Champaign
• Experimentalist at heart, focus on chassis
  dynamics, systems engineering, and control
• Service
• Chair of ASME Automotive and Transportation
  Systems Committee
• National Academies Transportation Visualization
  Committee
• Organizer for ASME DSCC conf, IEEE Conf on
  Control Applications
• Faculty advisor for Penn State Robotics Club,
  AUVSI Competition
• Teaching
• Department teaching award, 2006, College teaching
  award in 2007
• SAE Teetor award, 2008
• Research
• 3 million in ongoing research across 6 research
  labs
• Support 10 to 15 grad students, 10 undergrad
  researchers
• Selected as top papers at 07 IFAC Advances in
  Automotive Control
• Best paper in session, 2007 ASME IMECE

3
We do vehicle dynamics and control
Thats me doing a demonstration for my Vehicle
Dynamics course, Spring 2008!
See http//controlfreaks.mne.psu.edu for more info
4
Advanced estimation and virtual driving
See http//controlfreaks.mne.psu.edu for more info
5
Robotics and systems integration
See http//controlfreaks.mne.psu.edu for more info
6
Outline

• Vehicle dynamics
• Advanced estimation
• Robotics and systems

See http//controlfreaks.mne.psu.edu for more info
7
Vehicle Dynamics
High-speed ground robots
Passenger vehicle and hybrid vehicle control
Heavy Vehicle Reliability
4 wheel steering
Hybrid-electric military vehicles
Jack-knifing scale vehicles
See http//controlfreaks.mne.psu.edu for more info
8
Full-scale vehicle dynamics testingThe facility

• PTI Test Track
• One of a few closed-access University-owned
  test-track facilities
• Built to accommodate passenger and heavy
  vehicles,
• Only facility certified for bus chassis testing

9
Vehicle dynamics and model fitting

• looking under rocks

Vishi, Sittikorn, John, Bridget, Ryan, Dennis,

• Experimental testing
• We rarely trust other peoples models. Despite
  many claiming that they are rock solid, its
  the muddy fits that are of interest to us.
• As a consequence, nearly every student in my
  group is trained in vehicle dynamic validation
  and data collection

See http//controlfreaks.mne.psu.edu for more info
10
Model FittingFrequency Response Roll Angle
See http//controlfreaks.mne.psu.edu for more info
11
Model Fits Time Domainlane change
Yaw Rate
Roll
Lateral Velocity
Difficulty hard
Difficulty medium
Difficulty easy
Assuming terrain influence is removed more
later
See http//controlfreaks.mne.psu.edu for more info
12
Scale vehicle dynamics dangerous scenarios

• playing with toys

Sittikorn, Alexia, Andrew, Janine, Gareth

• For many instances of vehicle testing, the use of
  a full-sized vehicle is costly and dangerous, and
  yet simulations are onerous and questionable to
  build
• One solution often used is a reduced-scale
  vehicle.
• Mathematics of dimensional analysis allows
  results to map between behaviors of a scaled
• vehicle and those of a full-sized vehicle.

13
Some examples
15 scale wheel-lift characterization
15 scale Platooning dynamics
18 scale autonomous motorcycle
114 scale jacknifing
18 scale vision-tracking
14

• 15 scale
• Multi-input system
• Each axle is independently steered
• Each wheel has independent torque control

15
See http//controlfreaks.mne.psu.edu for more info
16
Comparisons between vehicles

• our hobby is collecting vehicle data

Sittikorn, Mariona, Haftay, Dennis, Jon

• Use same techniques as used in wind-tunnels,
  Buckingham Pi Theorem

17
Comparisons between vehicles

• our hobby is collecting vehicle data

Sittikorn, Mariona, Haftay, Dennis, Jon
from publications
from NHTSA database
Outlier data
18
Advanced estimation
Advanced sensor fusion
Redundant estimation
Vehicle-terrain interaction
Path of Lidar Sensor
Bridge with cement barriers on either side
See http//controlfreaks.mne.psu.edu for more info
19
The influence of terrain
Bridget

• Road grade investigated for steady state circle
  at various speeds
• When aligned based on path distance covered, the
  road grade measurement is very repetitive
  irregardless of speed

20
The influence of terrain
Bridget
Because feedback gains are directly related to
modeling error, disturbance cancellation enables
much higher gains and hence better tracking in
closed-loop control.
21
Terrain as a sensor

• GPS was never meant to be trusted for feedback
  control

Adam, Ryan, Vishi
Off-line Localization using Pearson Correlation
Coefficient By comparing pitch disturbances with
a terrain map, we are able to resolve
longitudinal position as good as 10 cm
22
Terrain as a sensor

• Learning as you go

Adam, Ryan, Vishi
Representative visualization of your work
Real-time Localization using Particle
Filters Tested again at the track we are able to
resolve longitudinal position to 0.5 meters after
traveling about 100 meters, with no GPS or other
signals
23
Mapping terrain
Pramod
The goal of this work is to map road features and
thereby correlate results to accident causation
and eventually prevention Impact 2000 lives
saved a year!
24
Mar 08
Oct 07
Nov 07
Apr 08
Dec 07
See http//controlfreaks.mne.psu.edu for more info
25
Mapping terrain

• Shown at right is a banked curve from the test
  track
• Getting 10 to 30 scans per second out to 80
  meters of range.
• Accuracy on the order of 6 cm at best case
  (perfect GPS).
• Actual error is on the order of a meter or less.

Path of Lidar Sensor
Asphalt Roadway
26
Example bridge section
Path of Lidar Sensor
Bridge with cement barriers on either side
Asphalt Roadway
See http//controlfreaks.mne.psu.edu for more info
27
See http//controlfreaks.mne.psu.edu for more info
28
Advanced sensor fusion how to utilize map-based
position?

• Can get orientation!
• Real and virtual scenes are compared.
• Preliminary results show orientation accuracies
  of 0.1 deg

Vishi, Adam
29
See http//controlfreaks.mne.psu.edu for more info
30
Automation and systems integration
4 wheel steering
High-speed ground robots
Hybrid-electric military vehicles
Autonomous vehicle testing
See http//controlfreaks.mne.psu.edu for more info
31
Solving automation challenges

• Want to measure driver steering torque and
  backlash effects caused by steering systems,
  suspension, tire behavior, etc.

• Problem need standardized interface to measure
  driver inputs to the steering system and hence
  tire
• Senior project?

32
Off-road modeling

• Preventing the accident in the first place

Bridget, Jason
Currently using Monte-Carlo methods and CarSim to
analyze the effect of highway geometry on
accident causation
See http//controlfreaks.mne.psu.edu for more info
33
Predicting and preventing unintended roadway
departure

• According to FHWA, 60 of vehicle fatalities
  occurred after leaving the lane
• High-gain control combined with terrain maps
  gives an unprecedented opportunity to mitigate
  this through the steering input

34
Efficiency improvements by sensor fusion

• anticipating the road ahead

Nan, Alexia, Vishi
See http//controlfreaks.mne.psu.edu for more info
35
HEMTT Starter SystemHIL Project

• Army is spending 30 million each month on
  premature battery failure
• Project ultracapacitor switchover
• More reliable starts
• Vastly increase battery life
• Adaptive to extreme environmental temperatures
• (HIL) Test Stand Simulator
• Simulates HEMTT engine, alternator, battery and
  ultra-capacitor
• Responds to inputs from actual HEMTT starter
  motor
• Records speeds and torques of starter motor and
  engine

See http//controlfreaks.mne.psu.edu for more info
36
Why is battery management necessary?
For manned vehicles, reliability is important,
but logistics and support costs are huge
For unmanned vehicles, logistics and support
costs are also important, but vehicle runtime and
operator safety are paramount (example EOD bots)
37
Campus-wide hardware-in-the-loop project
38
The goal of this work is to accelerate hybrid
vehicle powertrain development
Faculty Participants Dr. Sean N.
Brennan Lab/Center Name GATE Hardware-in-the-Loo
p Sponsor DOE

• Distributed Powertrain System
• Utilize campus-wide Ethernet
• Incorporate existing labs
• Integrate with industrial facilities

Chassis Dyno
IC Engine
Electric Motor
Ultracapacitor
Flywheel
Driving Simulator
Battery
Fuel Cell
See http//controlfreaks.mne.psu.edu for more info
39
Analyzing reliability of PTI bus testing results
Faculty Participants Prof. Sean
Brennan Lab/Center Name Pennsylvania
Transportation Institute Sponsor Federal Transit
Administration

• Future Work
• Develop a predictive failure model to aid transit
  agencies in making purchase decisions

• Collecting reliability data from transit agencies
  around USA
• Comparing transit agency data with PTI test track
  data to assess their validity

40
See http//controlfreaks.mne.psu.edu for more info
41
See http//controlfreaks.mne.psu.edu for more info
42
See http//controlfreaks.mne.psu.edu for more info
43
Allometric Design and Stability Relationships for
Explosive Ordinance Robots
Participants Brennan, Dean, Logan, Labs
Intelligent Vehicles and Systems Group, ARL,
EDOG Sponsor NAVEOD (DoD)

• Real-time Localization using Particle Filters
• By comparing inertial disturbances with a terrain
  map, we are able to resolve longitudinal position
  to 0.5 meters after traveling about 100 meters,
  with no GPS or other signals

• Read more
• Guizzo, Erico. 280 Million Robot Dustup,
  IEEE Spectrum, p. 10-13, Vol. 44, No. 12, North
  American Edition, December 2007.

44
New frontiers

• Recently initiated studies on human-vehicle
  interaction using a recently donated immersive
  driving simulator

45
New frontiers remote semi-autonomy and driver
assist
Nan, Alexia, Vishi
Use immersive driving simulator to remotely guide
vehicles through pre-mapped terrain.
46
Thanks to supporters!

• The National Science Foundation funded research
  into fundamentals of dynamic behavior through
  several student fellowships. (200k)
• The National Academy of Science, The
  Transportation Research Board funded roadway
  scanning and terrain modeling (300k)
• Army TACOM currently funding HIL work (1M)
• The Federal Transit Agency funded test track
  and vehicle systems used on the track such as the
  DGPS/IMU system (track 14M, current project
  300k)
• Naval Explosive Ordinance Disposal currently
  funding robotics work (600k)

47
Questions?

• Vehicle dynamics
• Advanced estimation
• Robotics and systems

See http//controlfreaks.mne.psu.edu for more info