Date post: | 30-Dec-2015 |
Category: |
Documents |
Upload: | oscar-lambert |
View: | 29 times |
Download: | 0 times |
University of California, Berkeley
Intelligent Vehicle-Highway Systems
Shankar Sastry
California PATH University of California, Berkeley
(Joint work with Datta Godbole, John Lygeros, Raja Sengupta & Shankar Sastry)
University of California, Berkeley
Intelligent Vehicle-Highway Systems (IVHS)
Partially or fully automate driving on the highways– can increase driving comfort and reduce stress– potential for increased safety
• 90% of all accidents are attributed to human error• Although many more hazards are successfully handled by humans.
– Automation can induce structured environment and tight control resulting in high capacity, less pollution & guaranteed travel times
Types of Automation – Driver Warning & Assistance (e.g., Blind Spot Warning)– Emergency Control (ABS,Daimler Chrysler schemes)– Control of Repetitive Tasks (Adaptive Cruise Control)– Complete Control (Automated Highway SystemsAutomated Highway Systems)
University of California, Berkeley
Control Problems in IVHS
Objectives– Increase safety & efficiency of the existing highway
infrastructure• objectives of the individual users and the system may not match
Characteristics– Control Design: Multiple Agents Compete for Scarce Resources
• Centralized control can yield optimal solutions but may be too complex and unreliable (danger of single point failure)
• Decentralized control increases reliability but may result in non-optimal or even unsafe solutions.
– Performance Evaluation• Performance metrics specified in terms of overall system whereas controllers
designed for individual vehicles• Evaluation in the uncertain environment of partial automation
University of California, Berkeley
Automated Highway System
Fully Automated Vehicles Operating on Dedicated Lanes– Involves control of individual vehicles as well as their
collective behavior Conflicting Objectives
– Safety & Capacity– Travel Time & Throughput (Individual vs System Optimal)
Definition of Safety– Ideally no collisions– Allowing low relative velocity collisions results in two
acceptable longitudinal vehicle following configurations• Following very close (platoon follower)• Following at sufficiently large distance (platoon leader)
University of California, Berkeley
Automated Platoons on I-15
University of California, Berkeley
Control of Automated Highway Systems
Design of vehicle controllers & performance estimation Two concepts
– platooning & individual vehicles
Network
Link
Coordination
Regulation •Lane keeping•Vehicle following
•Maneuver selection•inter-vehicle comm
•Dynamic routing
•Flow optimization Entry
Exit
LaneChange
PlatoonFollowing
Join
Split
Speed,vehiclefollowing
University of California, Berkeley
Vehicle Following & Lane Changing
Control actions: (vehicle i) -- braking, lane change Disturbances: (generated by neighboring vehicles) -- deceleration of the preceding vehicle
-- preceding vehicle colliding with the vehicle ahead of it
-- lane change resulting in a different preceding vehicles
-- appearance of an obstacle in front Operational conditions:
– state of vehicle i with respect to traffic
i
j
i-1 i-2
University of California, Berkeley
Game Theoretic Formulation
Requirements– Safety (no collision)
– Passenger Comfort
– Efficiency• trajectory tracking (depends on the maneuver)
Safe controller (J1): Solve a two-person zero-sum game
– saddle solution (u1*,d1*) given by• Both vehicles i and i-1 applying maximum braking• Both collisions occur at T=0 and with maximum impact
J x u d x t J Ct
10
03 1 1 0( , , ) inf ( );
J x u d u t J C mst
20
02 2
32 5( , , ) sup| ( )|; .
u U d D J x u d J x u d J x u d, , ( , , ) ( , , ) ( , , )* * * * * 10
1 10
1 1 10
1
University of California, Berkeley
Safe Vehicle Following Controller
Partition the state space into safe & unsafe sets0
min,304
02
01 ),,(: xxxxS
Design comfortable andefficient controllers inthe interior•IEEE TVT 11/94
Safe set characterizationalso provides sufficientconditions for lane change•CDC 97, CDC98
University of California, Berkeley
Automated Highway System Safety
Theorem 1: (Individual vehicle based AHS) – An individual vehicle based AHS can be designed to produce
no inter-vehicle collisions, – moreover disturbances attenuate along the vehicle string.
Theorem 2: (Platoon based AHS)– Assuming that platoon follower operation does not result in
any collisions even with a possible inter-platoon collision during join/split, a platoon based AHS can be safe under low relative velocity collision criterion.
References– Lygeros, Godbole, Sastry, IEEE TAC, April 1998– Godbole, Lygeros, IEEE TVT, Nov. 1994
University of California, Berkeley
Estimate maximum per lane capacity as a function of– vehicle braking rates, delays, types of coordination
Individual vehicles can increase highway capacity by a factor of two:– on-line estimation of braking capability
Platooning provides similar capacity with the possibility of low impact velocity collisions– Consider: emergency deceleration for obstacle avoidance
• differences in delays & braking rates give rise to multiple and severe intra-platoon collisions requiring larger separation between two platoons
References– Carbaugh, Godbole, Sengupta, Transportation Research-C, 98– Godbole, Lygeros, Transportation Research-C, 99
AHS Performance Evaluation
University of California, Berkeley
Highway Capacity Estimate (Single-Lane)
Queuing Analysis
•Up to 20% capacity loss due to entry and exit•Up to 15% loss due to lane changes•Platoon Join/Split ??
N=Platoon size
References•Transportation Research part-C: 1998, 1999
University of California, Berkeley
Fault Management
Faults induce switching of control strategies at multiple levels of hierarchy to maintain safety and minimize performance degradation
Design of fault management system– fault identification (distributed observation)– fault classification– fault handling
• minimal set of new maneuvers• fault localization• verified logical correctness of communication protocols
Need for probabilistic verification– worst-case design can not produce a safe system with faults
– given component reliability & Pd-fa characteristic of fault identification algorithms, compute probability of collisions.
University of California, Berkeley
AHS Control Architecture
Network
Link
Coordination
Regulation •Multi-ObjectiveControl Design
•Safe & efficientControl Switching•Inter-vehicle comm
•Dynamic routing
•Flow optimization
Network
Link
Coordination
Regulation
•Flow optimization
•Multi-ObjectiveControl Design
•Safe & efficientControl Switching•Inter-vehicle comm
•Dynamic routing
Network
Link
Coordination
Regulation •Multi-ObjectiveControl Design
•Safe & efficientControl Switching•Inter-vehicle comm
•Dynamic routing
•Flow optimization
Fault Mode i
Fault Mode j
AnalysisMethodsandTools
OperatingScenario
SystemPerformance
University of California, Berkeley
Deployment of AHS
Partial Automation yields progressive deployment path– Lack of structured environment– Lack of the knowledge of other driver’s intentions– Greedy driving policies– Human factors issues are highly pronounced
• false alarms, nuisance alarms, driver attentiveness, risk compensation, role confusion (Godbole et. al. TRB 98; James Kuchar at MIT)
Designing concepts for partial automation – ACC only roadway with infrastructure assisted entry
• (Godbole et. al. TRB 99)
Benefit Evaluation of partial automation systems– Hierarchical benefit evaluation methodology that integrates
analysis, simulation and experimentation results• adopted by NHTSA for crash avoidance systems analysis at VOLPE labs