Cooperative maneuvering in road traffic

Transcription

1 Cooperative maneuvering in road traffic ISN Conference 2014 Jeroen Ploeg

2 Outline Introduction & motivation Cooperative Adaptive Cruise Control Cooperative maneuvering Design approach Interaction protocol Vehicle control system architecture Test vehicle implementation Test results Conclusions 2

3 Introduction Societal demand for more efficient, cleaner, and safer road traffic Expand transportation infrastructure Fuel-efficient and safe vehicles Automation Cooperative Driving Main enabler: wireless communication Longitudinal vehicle automation Adaptive Cruise Control: time gap > 1 s Cooperative Adaptive Cruise Control (CACC) short inter-vehicle distance: time gap 0.3 s increased road capacity decreased fuel consumption 3

4 CACC control objectives a ref,4 a ref,3 a ref,2 a ref,1 5 v 5 v d 5 d 4 v 3 d 3 v 2 d 2 v 1 Control objectives Vehicle following d i d r,i desired distance: d r,i = r + hv i String stability Attenuation of the effect of disturbances over the vehicle string v i v i string unstable t string stable 4 t

5 Cooperative maneuvering What about platooning under common highway scenarios? Leaving, entering Lane extension, reduction Scenario execution requires a well-defined interaction protocol. Interaction protocol is characterized by A sequence of maneuvers to execute a scenario A corresponding wireless message sequence to inform the vehicles that are engaged in the scenario execution Hence, the vehicle CACC system must include functionality to execute such an interaction protocol 5

6 Scenario decomposition: Maneuvers To obtain a generic solution, scenarios are decomposed into maneuvers Examples scenario lane reduction maneuvers gap making merging platooning scenario lane extension maneuvers splitting platooning 6

7 Maneuver execution: Agents The maneuvers are executed by a set of high-level controllers, socalled agents Multiple agents may be running at the same time to execute a maneuver Examples gap making merging maneuvers CACC CACC obstacle avoiding obstacle avoiding agents vehicle/lane following lane changing 7

8 Agent-based control Design is based on defining different agents for different functionalities α agent (cooperative ACC CACC) β agent (obstacle avoiding OA) γ agent (lane changing LC) agent (vehicle/lane following V/LF) Each agent contributes in the control action through a separate term u i = f i α + f i β + fi γ + fi δ + 8

9 Scenario example: Lane reduction Maneuvers and active agents lane reduction gap making merging platooning CACC obstacle avoiding lane changing vehicle/lane following 9

10 Not yet? Not yet? Not yet? Not yet? Not yet? Interaction protocol Follow vehicle 1 (CACC, V/LF) 3 CACC, V/LF CACC, V/LF 1 Increase distance to 2 (OA) 2 Far enough from 2? Tune to platoon speed (CACC) Tuned! Far enough from 1? (CACC) Send STOM Merge request Merge request Increase distance to 1 (OA) Far enough from 1? Safe-TO-Merge (STOM) Yes! Wait for STOM Agents CACC: Cooperative ACC OA: obstacle avoiding LC: lane changing V/LF: vehicle/lane following 10 Yes! Do merge (LC)

11 Control system architecture Supervisory layer Control layer Perception / information layer 11

12 Information layer Tracked objects for CACC Additionally tracked objects for cooperative maneuvering 5 3c 1 2 host 3b 4 3a a, b, c: only one at a time 12

13 Test vehicle implementation Implementation in 3 cars (Toyota Prius III Executive) Forward-looking radar Forward-looking camera ITS G5 (IEEE p) wireless communication 13

14 Test results: Split & merge maneuver 14

15 Test results: Lateral vehicle control Lateral control modes Lane keeping Vehicle following Modes included in Vehicle/Lane Following (V/LF) agent Merging involves multiple mode-switching events 15

16 Conclusion Structured agent-based approach to scenario implementation Decomposition of a scenarios into multiple maneuvers Each maneuver executed by one or more agents Distributed solution, i.e., no central intelligence needed, while minimizing the communication overhead, without the need to have a platoon leader/supervisor, such that a single solution (agent) might serve several objectives (maneuvers). Solution would also work with non-equipped merging vehicle (using direction light detection) Hence, cooperative platooning will be capable of dealing with common highway maneuvers 16

17 Cooperative maneuvering in road traffic ISN Conference