Designing an Automated Vehicle System

1
Designing an Automated Vehicle System

• Raja Sengupta
• Assistant Professor
• Civil and Environment Engineering
• University of California, Berkeley

2
The System
Click URL Below to View Media Clip
http//www.path.berkeley.edu/CCIT_BART_Symposium/d
emo_97.wmv
3
How does it work?Control

• Distributed control
• Linear function of leader acceleration, speed,
  front vehicle position, acceleration, speed
• String Stable

Cite Hedrick, Swaroop etal
4
How does it work? Hardware
5
How does networking work?Our Wireless Token Ring
Protocol
T
Data
PS (C) B NS (C) A
C
T
Data
Implemented on 802.11b WiFi
T
Data
B
PS (B) A NS (B) C
A
PS (A) C NS (A) B
A Wireless Token Ring Protocol for Intelligent
Transportation Systems. IEEE ITSC, August 25-29,
2001.
6
How does it work?

• Feedback control design
• Methodology for an adaptive cruise control study
  using the SHIFT/Smart-AHS framework. IEEE SMC
  1998.
• A method for design and specification of
  longitudinal controllers for vehicle automation,
  ITS America. 1998
• Rear-end crash mitigation benefits of an
  automated highway system. SAE FTT98.
• Supervisory logic and maneuver coordination
  design
• Distributed hybrid controls for automated vehicle
  lane changes. IEEE CDC 1998.
• Design of emergency manoeuvres for automated
  highway system obstacle avoidance problem. IEEE
  CDC 1997
• Monitoring system design
• Fault diagnosis for intra-platoon communications.
  IEEE CDC 1999.
• A methodology for the integration of vehicle
  failure diagnostics. ITS America 1998.
• Wireless Network design
• Tools for the design of fault management systems.
  IEEE ITSC 1997
• Tools for safety analysis of vehicle automation
  systems. ACC 1997

7
Our Design Process

• Prove the elements and their integration are
  correct whenever possible
• Correct safe/fair/stable/optimal
• Usually done in models that leave out a lot of
  real world detail
• Simulate to learn more
• Use models that are closer to the real world
• Calibrated/validated.
• Experimental testing to learn still more
• Produce real-time software, put it on the
  hardware
• Ideally prototype production process should
  preserve all the properties
• Those proven, established by simulation,
  established experimentally
• This is difficult Mars Rover, Three mile island,
  .
• Established Techniques Fault tree, root locus,
  hatley-pirby diagrams, ..
• Research Techniques Hybrid specification,
  analysis, code generation,

8
Proving Correctness The box example
9
(No Transcript)
10
Proving Correctness The Box Example
11
Similar Correctness Proofs can be generated for
carsAutomated Merge Example
12
The Logic And The Continuous Dynamics Are Brought
Together In Hybrid Specifications
13
Proving Correctness of Hybrid Specifications

• Typically safety is proved in a
• simplified generalization of the
• real world

14
Hybrid Specifications Can be SimulatedDoes our
control design reduce shockwaves?
B
C
A
Merge-in Point

• The head vehicle A in the merge-in lane
  broadcasts a message to all main lane vehicles
• The Irrelevant vehicle B ignores the message
• The relevant vehicle C in the main lane brakes
  upon receipt of the message, making a proper gap
  in advance
• Vehicle A merges in after t seconds if he sees a
  good gap in the main lane, otherwise waits at
  merge-in point
• If the main lane traffic is dense, a queue is
  formed in the merge-in lane

15
Trajectories of Main Lane Vehicles
ACC
CACC
16
Hybrid Specifications Can Be Converted Into
Real-time Software And Tested Token Ring Protocol
Top level Teja Spec of the Token Ring Protocol
17
Token Ring Protocol Test Data
3 FTP Flow
Token Rotation Time
Rotation number
18
Integrating Proofs, Simulation, and Software in
the Prototype Development Process
Syntax
Semantics
Proofs/ mathematical meaning
Code generation/ Compilation/ for Testing/ Simulat
ion
Approximates
Environment (car, computer, )

• Precision about these helps preserve proven,
  simulated, and tested properties in the prototype

19
An Example of Syntax and Semantics for Hybrid
Specifications
20
(No Transcript)
21
Precision Enables Tools To Be Used To Maintain
The Integrity Of Prototype Production
Syntax
Semantics
KRONOS, VeriSHIFT, Isabelle, MATLAB
SHIFT, MATLAB
Environment (desktop computer)
22
Precision Enables Tools To Be Used To Maintain
The Integrity Of Prototype Production
Syntax
Semantics
KRONOS, VeriSHIFT, Isabelle, MATLAB
Teja, Real-time Workshop
Environment
23
Conclusions

• Research in networked multi-vehicle systems
• Automated cars
• Unmanned air vehicles
• Technologies
• Wireless network design
• Distributed control design
• System safety and capacity design
• Use of design tools
• Hybrid specification, proofs, simulation, code
  generation
• Teaching CE 290I Control and Information
  Management
• Design and development of distributed real-time
  systems
• Study some part of BART?

24
Conclusions

• Technical Advisor to ASTM Committee for Dedicated
  Short Range Communications on Vehicle-Vehicle
  Communications
• Current partnerships
• FHWA
• PATH/CALTRANS
• Daimler-Chrysler Research
• General Motors Research
• Honeywell
• National Science Foundation
• Office of Naval Research
• DARPA