Thiết lập điều khiển cho xe tự hành
Được đăng lên bởi
vantuan912
Số trang: 6 trang

Lượt xem: 321 lần

Lượt tải: 0 lần
Accepted for publication in the 2012 IEEE International Conference on Robotics and Automation (ICRA 2012) Setpoint Scheduling for Autonomous Vehicle Controllers TszChiu Au, Michael Quinlan, and Peter Stone Abstract— This paper considers the problem of controlling an autonomous vehicle to arrive at a specific position on a road at a given time and velocity. This ability is particularly useful for a recently introduced autonomous intersection management protocol, called AIM, which has been shown to lead to lower delays than traffic signals and stop signs. Specifically, we introduce a setpoint scheduling algorithm for generating setpoints for the PID controllers for the brake and throttle actuators of an autonomous vehicle. The algorithm constructs a feasible setpoint schedule such that the vehicle arrives at the position at the correct time and velocity. Our experimental results show that the algorithm outperforms a heuristicbased setpoint scheduler that does not provide any guarantee about the arrival time and velocity. I. INTRODUCTION Recent developments in robotic vehicles lead us to believe that fully autonomous vehicles will be widely adopted in the future. Looking ahead to the time when such autonomous cars will be common, Dresner and Stone proposed a new intersection control protocol called Autonomous Intersection Management (AIM) and showed that by leveraging the capacities of such autonomous vehicles we can devise a reservationbased intersection control protocol that is much more efficient than traffic signals and stop signs [1]. The protocol, however, relies on the assumption that autonomous vehicles can always arrive at the intersection at a specific time and a specific velocity—if a vehicle enters the intersection at a different time and velocity, collisions may occur. To compensate for the sensing and control errors that could cause a vehicle to fail to meet that requirement, the intersection manager allows each vehicle to have a buffer— an area around the vehicle that no other vehicle can enter at any time in the intersection, such that even if the vehicles deviate from their given arrival times and arrival velocities, there is enough room to avoid collisions. The problem is that the buffer cannot be too large; if it is, the vehicle will need a lot of space in the intersection and will prevent other vehicles from using the space, causing a tremendous decrease in the efficiency of the protocol, as demonstrated by the mixed reality simulati...
Setpoint Scheduling for Autonomous Vehicle Controllers
TszChiu Au, Michael Quinlan, and Peter Stone
Abstract— This paper considers the problem of controlling
an autonomous vehicle to arrive at a speciﬁc position on a
road at a given time and velocity. This ability is particularly
useful for a recently introduced autonomous intersection man
agement protocol, called AIM, which has been shown to lead
to lower delays than trafﬁc signals and stop signs. Speciﬁcally,
we introduce a setpoint scheduling algorithm for generating
setpoints for the PID controllers for the brake and throttle
actuators of an autonomous vehicle. The algorithm constructs
a feasible setpoint schedule such that the vehicle arrives at
the position at the correct time and velocity. Our experimental
results show that the algorithm outperforms a heuristicbased
setpoint scheduler that does not provide any guarantee about
the arrival time and velocity.
I. INTRODUCTION
Recent developments in robotic vehicles lead us to believe
that fully autonomous vehicles will be widely adopted in the
future. Looking ahead to the time when such autonomous
cars will be common, Dresner and Stone proposed a new
intersection control protocol called Autonomous Intersection
Management (AIM) and showed that by leveraging the
capacities of such autonomous vehicles we can devise a
reservationbased intersection control protocol that is much
more efﬁcient than trafﬁc signals and stop signs [1]. The
protocol, however, relies on the assumption that autonomous
vehicles can always arrive at the intersection at a speciﬁc
time and a speciﬁc velocity—if a vehicle enters the intersec
tion at a different time and velocity, collisions may occur.
To compensate for the sensing and control errors that
could cause a vehicle to fail to meet that requirement, the
intersection manager allows each vehicle to have a buffer—
an area around the vehicle that no other vehicle can enter at
any time in the intersection, such that even if the vehicles
deviate from their given arrival times and arrival velocities,
there is enough room to avoid collisions. The problem is that
the buffer cannot be too large; if it is, the vehicle will need a
lot of space in the intersection and will prevent other vehicles
from using the space, causing a tremendous decrease in the
efﬁciency of the protocol, as demonstrated by the mixed
reality simulation conducted by Quinlan, et al. [2].
Au et al. developed a motion planning algorithm for
autonomous vehicles to arrive at the intersection at a speciﬁc
time and velocity [3], thus allowing a much smaller buffer
size. The motion planning algorithm, however, is based on a
mathematical model of vehicle control that is too simplistic
when compared with the control of a real vehicle. For
example, the algorithm assumes vehicles can maintain a
linear acceleration until arriving at a given velocity, but
Department of Computer Science, The University of Texas at Austin.
{chiu,mquinlan,pstone}@cs.utexas.edu
Fig. 1. The architecture of the controller for the brake and throttle actuators.
the acceleration can be far from linear if the vehicles are
controlled by PID controllers, especially when starting from
a stationary position or decelerating after a sharp brake.
Therefore, the algorithm does not achieve its intended effects
on a real autonomous vehicle.
In this paper, we present a new motion planning algorithm
called a setpoint scheduler that is based on a more realistic
model of vehicle control. The model is built via empirical
performance proﬁling of the PID controllers for the brake
and throttle actuators of a vehicle. In addition, we develop a
smoothing technique for computing a sequence of setpoints
that allows the vehicle to slow down gracefully without
hitting the brake too hard. The setpoint scheduler searches
for a feasible trajectory for the vehicle based on a descriptive
model of the PID controllers’ performance. We implemented
the setpoint scheduler on an autonomous vehicle, and exper
imentally compare it with the vehicle’s default PIDbased
reactive controller that heuristically computes the setpoints
based on the given arrival time and velocity.
II. MODELING VEHICLE PERFORMANCE
Our setpoint scheduler is designed for controlling the
Austin Robot Technology vehicle, an Isuzu VehiCross that
has been upgraded to run autonomously [4]. Fig. 1 shows
the architecture of the controller of the brake and throttle
actuators. The setpoint scheduler takes the measures from
the odometer and the speedometer and computes the target
velocity as the setpoint for the PID controllers which control
the positions of the brake and throttle.
Our setpoint scheduler needs to know the effect of setting
a setpoint in order to predict the movement of the vehicle.
Given the current velocity v and a target velocity ˆv, the
scheduler needs to know how long the PID controllers will
take to stabilize the velocity of the vehicle at ˆv after setting
the setpoint to ˆv, and how much the vehicle will move
before its velocity is stabilized. Thus our approach relies
on the estimation of two functions T
stable
and D
stable
, where
T
stable
(v, ˆv) is the longest time the vehicle takes to stabilize at
ˆv and D
stable
(v, ˆv) is the average distance the vehicle travels
after setting the setpoint to ˆv for a period of T
stable
(v, ˆv). We
Accepted for publication in the 2012 IEEE International
Conference on Robotics and Automation (ICRA 2012)
Để xem tài liệu đầy đủ. Xin vui lòng
Đăng nhập
Nếu xem trực tuyến bị lỗi, bạn có thể tải về máy để xem.
Thiết lập điều khiển cho xe tự hành

Người đăng:
vantuan912
5
Tài liệu rất hay!
Được đăng lên bởi
trinhhao162

1 giờ trước
Đúng là cái mình đang tìm. Rất hay và bổ ích. Cảm ơn bạn!
6
Vietnamese
Thiết lập điều khiển cho xe tự hành
9
10
266