1. Car Periphery Supervision System: Case Study 3
Just the first page
Ed Brinksma
University of Twente

2. CPS: Informal description
CPS obtains and makes available for other systems information about environment of a car. This information may be used for:
Parking assistance
Pre-crash detection
Blind spot supervision
Lane change assistance
Stop & go
Etc
Based on Short Range Radar (SRR) technology
The CPS considered in this case study
One sensor group only
(currently 2 sensors)
Only the front sensors and corresponding controllers
Application: pre-crash detection, parking assistance, stop & go

3. CPS: General model timing analysis Hybrid Stochastic Car Environment
Sensors
Measurement control (ECU)
Stochastic
ECU – tasks
on OSEK
Car
Airbag control
Belt Tensioner
HMI
Etc
timing analysis
Sensor
CAR
ECU
Tasks
running on
OSEK
HYBRID
STOCHASTIC
TIMED W O R L D

4. Requirements, assumptions & constraints
Proposed requirements (initially unclear)
To provide the information about the environment to the belt tensioner, the air bag controller and HMI on time
Avoid false alarms, i.e. to provide reliable information
Proposed assumptions & constraints
Environment (physical) constraints – a maximal (de-) acceleration, a maximal velocity, etc.
Information consumers constraints – activation time, etc.
Sensors properties – placement, physical properties
ECU constraints – computation delays, communication delays, etc.
Requirements, Assumptions and Constraint – why and what

5. ? Sensor scan modes 10 m DScan mode cycle: appr. 10 ms
1.4 m CVScan mode
cycle: < 1.5 ms
2 m DScan mode
cycle: appr. 2 ms

6. Environment Analysis Different width of cars (1.5-2m)
Interval for the sensors placement ( m)
Interval of velocities (5-60m/s)
Visibility analysis (MatLab)
difference is big, but not important for the current services
Trajectories analysis
CVMode information is important only for the Air Bag Controller
For the Belt Tensioner only early DScan information is important

7. UPPAAL model Env 1 Env 2 Sensor 1 Sensor 2 Measurement controller ECU
Periodic
scan
Mode switch
d=f(d1,d2) d1 d2
Env1,Env2: distance between object and car
The front of the car is divided into 12 smaller regions
Sensor mode is switched from long distance scan to short distance scan based on “d”
ECU’s knowledge about the object in collision course is at most one region behind
Under assumption on the velocity of the object and the car, and the number of objects, the property can be satisfied

8. Results Requirements, assumptions & constraints list
Meeting in Frankfurt, Robert Bosch GmbH (03/2003)
Visibility analysis
Trajectories analysis
CVScan information is important only to the air bag controller
CVScan and DScan (2m) is not relevant for the belt tensioner (if velocity is more than 20m/s~65km/h)
Short DScan may be tuned to the air bag controller or the belt tensioner needs
Presentations in Ametist meetings in
Twente (09/2002)
Dortmund (11/2002)
Cassis (4/2003)
Conferences
B.Gebremichael, H.Hermanns, T.Krilavičius, Y.S.Usenko Hybrid Modeling of a Vehicle Surveillance System with Real-Time Data Processing, Dynamical System Modelling and Stability Investigation, Kiev, 05/2003
Technical results – left side, Activities – right side
Uppaal model, verification of certain properties

9. Bosch has put the SRR system development on ice.
Plans
Recent news:
Bosch has put the SRR system development on ice.
Wider set of the trajectories
with (de-) acceleration
non-orthogonal trajectories
Extension of problem
hybrid analysis
control aspects
active cruise control
platoon navigation
Applications of the modeling results to tune the Uppaal model
Pushing tools to the limit and using state of the art methods to overcome these limitations
Technical plans – the left side and the bottom right side, General – the top two on the left side
Experimentation with different short DScan distances
Tuning the Uppaal model to the tool