The Possibility of Stress Free Driving

1
The Possibility of Stress Free Driving

• By Petersen W. Gross

2
Summary

• Introduction
• Sensors
• Obstacle Detection
• Route Planning
• Data Processing
• DARPA Grand Challenge
• Conclusion

3
Introduction

• Vehicles has changed greatly
• How they start
• Comforts
• And more
• Driving is not an easy task
• Requires 3 of 5 senses
• Sight
• Hearing
• Touch
• Need a license

4
Sensors

• Many different types
• Detect different objects
• Detect lane
• Detect people
• Detect stop lights
• And more!
• Determine vehicles location

5
Sensors of Team TerraMax

• Two GPS units
• Gyroscope
• Magnetic compass
• Four laser rangefinders
• Two radar units
• Twelve ultra sonic rangefinders
• Two pairs of CCD digital color cameras

6
Sensors of SciAutonics

• GPS
• Gyroscope
• Ladar
• Optical Sensors
• Smaller vehicle, less sensors

7
Other Solutions

• Detecting obstacles in front only
• Single front mounted camera
• Driver Assistance
• Panning ultrasonic sensor
• Panning camera
• Inclinometers

8
Why all the Sensors?

• GPS need for vehicle location
• Gyroscope and compass used when no GPS
• Cameras detecting surroundings
• Radar/ladar finding obstacles

9
Problems with Sensors

• Life time varies
• Weather
• With stand a beating

10
Needed Sensor Improvements

• Less Noise
• Smaller size
• Vehicle placement

11
Obstacle Detection

• One of the hardest parts
• Obstacles very
• Height
• Width
• Shape
• Size
• Lane detection

12
Detection of The Road and Lanes

• Road designed in predictable layout
• Lane tracking relatively easy
• Two white lines
• White line and double yellow

13
Intelligent Stop Go

• Two major objectives
• Automatic road following
• Lateral-control behavior
• Automatic road following
• Detect lead vehicle and follow
• Lead vehicle changes lanes, search for new
  vehicle
• Lateral-Control
• When following corners were cut
• Fixed by figuring out lane position

14
Intelligent Stop Go

• Detect pedestrians
• Shape
• How they walk
• First approach blurs outline of pedestrian and
  does principal-component analysis
• Second approached used motion detection and
  neural networks

15
MarVEye Project

• Tried to solve simple question of Where
• Ego centered view of location
• Lane position could be determined by using
  trigonometric functions

16
Team TerraMax

• Three Algorithms to determine lane and path
• 1st Textures of path and off-road will be
  different
• 2nd boundary detection using genetic algorithms
• 3rd multi degree-of-freedom

17
Route Planning

• Human drivers preplan a route
• Route can change
• Route planning also involves determining velocity

18
AutoNav

• Used a 4d approach
• Several different modules used
• Scene manager
• Deals with message passing
• Each other module must communicate with
• Gate-control
• Pan and tilt of cameras
• Gives back pan and tilt locations
• Vehicle control
• Access actuators and sensors of vehicle

19
AutoNav Continued

• VFE Module
• Communicates with VFE 100 (does image processing)
• Forward desired areas of interest to VFE hardware
• 3DS estimator
• Takes all data from other sensors
• Fuses data with image data
• Updates map of terrain

20
Velocity Planning

• Factors to determine velocity
• Weather
• Condition of road/terrain
• Other vehicles
• Time
• Distance
• Speed limit signs

21
Speed Limit Signs
22
Data Processing

• Great amount of data
• Each sensors data will not require as much CPU as
  next
• Image processing most CPU intensive
• Done with computers onboard

23
Some Examples

• Intelligent Stop Go
• Three 200 MHz Power PCs for city driving
• MarVEye
• Cluster of three PCs for image processing
• Separate system for everything else
• Team TerraMax
• Six computers (Pentium 4)
• Main control
• Map and route planning
• Sensor interface management
• Two for image processing
• Health and fault monitoring

24
Hardware Advances

• Power has increased greatly
• Leads to ability to process data faster
• Size of fully system has decreased
• Able to include more systems for data processing
• More systems, more sensors?

25
DARPA Grand Challenge

• Challenged issued by DARPA
• Two dozen teams met all requirements
• Couple of deadlines
• Video showing vehicle can operate w/o driver
• Visit by DARPA
• Navigate small course
• Course was about 150 miles
• Terrain varied
• 7 miles was furthest a vehicle went

26
Final Thoughts

• Proven possible
• Vehicle Design
• More research still needed
• The future of vehicles is autonomous
• Helps to save lives
• Normal drivers
• Armed Forces
• New products
• Autonomous lawn mower
• Autonomous boat