1 layout

5 downloads 0 Views 277KB Size Report
C o n tro lle r. Speed, Heading ... A c tu a to r. V is io n F ... M is s io n C o n tro l. Fig. 1. .... There are six VCM states: Standby ... sponsiveness, multi-tasking programming tech- ... It is written in ANSI C++ .... Edition), John Wiley, London, (1993).
STR/03/038/MECH

Development of Speed and Heading Control System for a Large Tracked Vehicle 1

1

Z. Gong, J. I. Guzman, S. J. Scheding , D. C. Rye , 1 1 G. Dissanayake and H. Durrant-Whyte obstacle avoidance and preferred road selection. Fig. 1 shows the system block diagram of the AGVS. The AGVS can also run with teleoperation control, with the systems for visual guidance and navigation being replaced by a tele-operation control system. More details on the AGVS can be found in [2-6].

Abstract – This report presents an innovative and effective algorithm for speed and heading control of a large tracked vehicle, which is retrofitted and developed to be computer controllable from a manual driving vehicle. Heuristic rulebased switching and adaptive PID control methods are used in this algorithm. With the implemented algorithm the vehicle control system is able to control the vehicle to operate on equatorial jungle-like natural terrain. A large number of field test data shows that the algorithm is robust and effective with excellent performances under various terrain conditions.

The tracked vehicle is originally designed for manual driving. It is a highly complicated nonlinear and uncertain system [7-8], which involves many mechanical systems, such as engine and its control system, gear and transmission, differential, track/soil interaction, and brake systems. Many attempts have been made to address various issues for the control of tracked vehicles [9-11]. However, there are very few reports on physical implementation of control systems for tracked vehicles.

Keywords: Autonomous ground vehicle system, Vehicle speed and heading control, Tracked vehicle, Switching control, Adaptive PID control 1

BACKGROUND

As part of the AGVS, the vehicle control system consists of the controller, the actuators, the speed and heading feedback units, and the tracked vehicle. This report describes the control system and an innovative and effective algorithm implemented for the speed and heading control of the tracked vehicle. Heuristic rulebased switching and adaptive PID control methods are used in this algorithm. A large number of field test data shows that the algorithm is robust and effective with excellent performances under various conditions.

In recent years, there has been great interest world-wide in development of autonomous ground vehicle system (AGVS) technologies, due to their immediate great potential in civil and military applications [1]. Sponsored by a government agency, Singapore Institute of Manufacturing Technology (SIMTech) embarked on a multimillion-dollar project to develop core AGVS technologies and implemented them on a large tracked vehicle. The project has been completed successfully and the developed AGVS is able to run autonomously on unstructured natural tropicforest-like terrain with the capabilities such as

M i s s io n C o n t r o l

V is u a l G u id a n c e

N a v ig a t io n

C o n t r o lle r

A c tu a to r

S p e e d , H e a d in g a n d P o s it i o n F e e d b a c k

V is io n F e e d b a c k

Fig. 1. System block-diagram of AUGVS. 1

Australia Centre for Field Robotics, The University of Sydney

1

V e h ic le

Development of Speed and Heading Control System for a Large Tracked Vehicle

2

OBJECTIVE

3.2

The objective of the project is to develop a practical and effective vehicle control system, which, together with other modules of the AGVS, controls the speed and heading of the tracked vehicle with required responses. The following works are involved in the project:

To develop the tracked vehicle into an autonomous ground vehicle, the drive-by-wire capability of the vehicle has to be implemented first. Imitating manual-driving operations, three servo systems together with specially designed actuation mechanisms are installed inside the driver compartment, to actuate the accelerator penal, and the left and right brake levers (see Fig. 2). A particular functional requirement is that whatever changes made in the driver compartment, this should still allow manual driving and the bypassing of the added servo actuating mechanisms.

To develop and implement algorithms for the vehicle speed and heading control. To develop software and hardware that interface with other modules of the AGVS, including those for navigation, actuation, and speed and heading feedback. 3

METHODOLOGY

3.1

The tracked vehicle

Actuation for computer controlled driving

Designed for manually driving, the tracked vehicle is powered by a diesel engine and, using skid steering method, steered through two brakes inside a controlled differential. The brakes are actuated mechanically by the driver via two hand-commanded levers. The vehicle transmission is semi-automatic. The gear ranges are selected manually with gear settings shifted automatically by the transmission. Some technical data of the vehicle is given in Table 1. Fig. 2. Actuators developed for computer controlled driving.

Table 1. Vehicle technical data.

Item Weight: Length: Width: Height: Max Land Speed: Steepest Grade: Engine Type:

Data 10,900 kg 4,851 mm 2,686 mm 2,705 mm 34 km/h(Range 1-2) 60% Two-stroke diesel

3.2.1

Compensation of brake mechanism

non-linearity

of

The driving mechanisms of the tracked vehicle are highly non-linear [7-8]. To achieve required control system responses, it is desirable to identify and compensate those non-linearities. Fig. 3 shows a large number of test data of the brake lever positions and the braking forces obtained in experimental characterisation trials. There is a clear dead band at the beginning positions of the brakes. It is also found that the brake force profile changes after vehicle maintenance. To compensate the non-linearity, the following smooth exponential-like function with three segments is used to represent the relationship between the brake lever positions and the braking forces:

There are three main items to operate for manual driving, i.e. an accelerator panel and left and right brake levers. Pressing/releasing the accelerator panel increases/decreases the engine power output and therefore the vehicle speed; pulling the left brake lever steers the vehicle to left; pulling the right brake lever steers the vehicle right; and pulling both the left and right brake levers reduces the vehicle speed. When driving, especially when turning by applying the brakes for skid steering, due to the coupling between the steering and braking, the coordination in actuating the accelerator and the two brakes is important for control of the vehicle speed and heading.

f = 0,

2

x ≤ 0.3

f = e a ( x −1) ,

x ≥ 0.5

f = e −0.5a ( x − 0.3) / 0.2,

0.3 < x < 0.5

(1)

Development of Speed and Heading Control System for a Large Tracked Vehicle

where f is the estimated braking force in the percentage of its maximum, x is the percentage of the brake lever position, and a is a constant, which is chosen to be 6, to give a nice fitted curve as shown with the red line in Fig. 3.

Fig. 4. Test results of engine speed.

3.3

The purpose of the vehicle control system is to ensure the execution of the speed and heading commands under the computer control. Fig. 5 shows a detailed block diagram of the system, which has two feedback loops. The inner loop represents the servo-actuators that respond to the set-point commands from the vehicle control module (VCM). The servo-actuators interact with the vehicle via a series of actuating mechanisms that apply position commands to the accelerator pedal and the left and right brake levers.

Fig. 3. Test results of brake forces.

3.2.2

Compensation of non-linearity of accelerator mechanism

In order to control the vehicle speed, we need to control the power output from the engine, which is closely correlated with the engine speed. Fig. 4 shows the test data of the engine speed under different accelerator pedal positions. It is observed that when the accelerator reaches to a certain position, the engine speed becomes very sensitive to the change of the accelerator pedal position. Based on the test results, the following function with two linear segments is used to represent the relationship between the accelerator pedal positions and the engine power output:

p=b

y , 0 .5

y ≤ 0.5

y − 0 .5 , p = b + (1 − b) 0 .5

The VCM acts as a controller in the outer feedback loop for the vehicle speed and heading control. It receives the speed and heading commands from a navigation module. According to the designed control algorithm, the VCM generates, at an updating rate of 10 Hz, the set-point commands of the servo-actuators for the accelerator and the left and right brakes. A vehicle position module, which consists of an Inertial Measurement Unit (IMU) and a GPS, provides speed and heading feedback. Due to the nature of the terrain, the GPS signal can be occluded for long periods of time, in this case the vehicle speed and heading are estimated by measuring the speed of both vehicle tracks using encoders attached to each of the track sprockets.

(2)

y > 0.5

where p is the estimated engine power output in the percentage of its maximum, y is the percentage of the accelerator pedal position, and b, chosen to be 10%, is a constant. C om m and A c tu a to r P o s it io n s

C om m and S peed C om m and H e a d in g

VCM C o n tr o lle r

Vehicle control system

A c c e le r a tio n A c tu a to r A c tu a to r S e rv o C o n tr o lle r

L e ft B r a k e A c tu a to r

A c t u a t o rs P o s it io n s

V e h ic le S peed

T rack ed V e h ic le

R ig h t B r a k e A c tu a to r

Fig. 5. Block diagram of vehicle speed and heading control system.

3

V e h ic le H e a d in g

Development of Speed and Heading Control System for a Large Tracked Vehicle

3.4

Speed and heading control algorithm

List 1. Pseudo Codes of Vehicle Speed and Heading Control Algorithm

The dynamics of the large tracked vehicles, together with the actuation mechanical systems, are highly complicated and non-linear [7-8]. The linear and rotating dynamics of the vehicle are strongly coupled. The skid steering method adds additional complicity in controlling the vehicle speed and heading.

// at each control updating get command speed and heading get actual speed and heading if (command speed large than limit) truncate it if (command speed is zero) { set engine idle set full brake send control updating end of control updating } // speed controller if (speed is too slow or a bit too fast){ // regulate speed by accelerator only compute accelerator setpoint based onmodified PID control principles clip setpoint between 0% and 100% anti-windup of controller integrator set SpeedBrake to zero } else { // speed is too fast, need brake set accelerator setpoint zero anti-windup of controller integrator compute SpeedBrake based PID control clip SpeedBrake between 0% and 100% } // heading controller check heading error and alarm if too large compute adaptive heading controllerparameters based on vehicle speed compute SteeringBrake based onmodified PID control principles anti-windup of controller integrator if (SpeedBrake exceeds SteeringBrake) { compute left and right brake setpointsby compromising brakes for steeringand speed reduction } else {// SteeringBrake exceeds SpeedBrake set SteeringBrake to left or rightbrake setpoint only based on turndirection } if ( going backwards ) { swap left and right brake setpoints } actuation non-linearity compensation send control updating end of control updating

To achieve the objectives of the vehicle control system, a practical and effective multivariable feedback control algorithm is developed for the vehicle speed and heading control system. With the proven robustness and simplicity of PID controllers in industrial applications, the PID control principles are adopted. However, a number of innovative modifications are made to the conventional PID controllers. The new control algorithm is of the following features. Switching control based on heuristic logic control rules. Adaptive PID controller gains. Coordination in speed and heading control. Non-linearity compensations. List 1 shows pseudo codes of the implemented speed and heading control algorithm for the tracked vehicle. Further details on modifications to PID control is described in the following. Limit integration in speed control. The integration value of the controller is limited, to avoid a deep saturation of the integrator, which may cause system instability. Sustain integration in speed controller. When the vehicle speed is larger than the commanded one over a threshold, the speed control output is set to zero and brakes will apply, but the integrator of the speed controller will not reset. This will help to avoid over dropping of the vehicle speed when it approaches the commanded one. Heading error warning. When the vehicle heading error is over a limit, a warning is generated. The abnormal heading error is a reliable and easy indicator for the system “out of control” condition. Adaptive PID control gains in heading control. The controller’s gains are adapted based on the vehicle speed. Limit integration in heading control. Same as for speed control, the integration value of the controller is limited. Reset of integrator in heading control. The integration is used only for small heading adjustment. When the heading error is large, the integrated heading error is reset to avoid over steering.

3.5

Control system implementation

3.5.1

Software implementation

In implementation of VCM control software, VCM states are defined and implemented with a state machine. There are six VCM states: Standby, Ready, Working, Exception, Emergency and Shutdown. These states determine the behaviour of the running of VCM control software. The state transitions are triggered after receiving commands from MCM, or, in the cases of exceptions, triggered by VCM internal events. The state transition diagram is shown in Fig. 6.

4

Development of Speed and Heading Control System for a Large Tracked Vehicle

lected. Some components are with military standard. Major hardware of the VCM includes:

Initialize

Standby

Shutdown

Ready

Emergency

Working

Embedded single board computer: Ampro Computers Inc., Model LB-P5x, EBX form factor. CAN communication board: Softing GmbH, Model CAN-AC2-104, PC104 form factor. Encoder interface board, ACS-Tech80 Model 5912-2, PC104 form factor. Power conditioning board, Diamond Systems, Model HE-104, PC104 form factor. Special protective measures, such as vibration isolation and air ventilation, are taken to protect the major components when they are installed on the vehicle. The extensive field tests at the natural equatorial environment proved that those prevention measures for a reliable system operation are very effective.

Exception

Fig. 6. VCM states transition diagram.

In order to meet the requirement of system responsiveness, multi-tasking programming techniques are employed in the implementation of the control system. The VCM controller program is implemented in a single process that contains a number of concurrent threads:

4

After the speed and heading control algorithm was implemented in the AGVS, extensive field trials were conducted on target equatorial junglelike natural terrain. Samples of responses of the speed and heading control system recorded in the field trials are shown in Fig. 7. Fig. 7(a) shows step speed and heading responses. Fig. 7(b) shows the responses under teleoperation control. Those real field test data shows that the performance of the vehicle speed and heading control system is satisfactory.

The MAIN thread implements the state machine and co-ordinates activities within all the threads of VCM. It provides communications with other modules of AGVS and that between the threads of the VCM. The HEART BEAT thread monitors the threads of VCM and provides periodic software signals that assures other system modules the continued activities of all the threads in VCM. The CONTROL thread implements the vehicle speed and heading control algorithm and provides periodic control outputs updating. The ENCODER thread computes the vehicle’s left and right track speeds based on the inputs from the encoders attached to the track sprockets.

5

CONCLUSION

In conclusion, an automatic vehicle speed and heading control system of unique features for a highly mechanical large tracked vehicle has been developed. This was completed through vehicle characterisation, theoretical and simulation analysis, control algorithm development, system implementation, and followed by field testing and commissioning.

The VCM control software is targeted to the Windows NT operating system executing on an IBM-compatible PC. It is written in ANSI C++ and compiled using the Microsoft Visual C++ compiler. 3.5.2

RESULTS & DISCUSSION

Heuristic rule-based switching and adaptive PID control methods are used for the vehicle speed and heading control. The controller is able to switch among different suitable control loops based on vehicle operation conditions. With this unique multi-input-multi-output control algorithm, the vehicle control system is able to control the vehicle to operate on equatorial jungle-like natural terrain. A large volume of field test record shows that the control algorithm is effective and robust with excellent performances under various environment conditions.

Hardware implementation

Leveraging on the availability of advanced PC and networking technologies, we implemented a cost effective and reliable hardware system for the vehicle control. Considering the harsh environment inside the vehicle, rugged electronic components with good performance are se-

5

Development of Speed and Heading Control System for a Large Tracked Vehicle

cle Systems and Technology, SANDIA REPORT2001-3685, Sandia National Laboratories, USA, (2001). [2] H. Wang, X. Jian, J. Ibañez-Guzmán, R. Jarvis, T. Goh and C.W. Chan, "Real Time Obstacle Detection for AGV Navigation Using Multi-baseline Stereo", ISER2000, Hawaii, USA, (2000). [3] P. Chaturvedi, E. Sung, A.A. Malcolm and J. Ibañez-Guzmán, "Real-Time Identification of Driveable Areas in a Semi-Structured Terrain for an Autonomous Ground Vehicle, Unmanned Ground Vehicle”, Aerosense 2001, Orlando, Florida, USA, 16-20 April 2001. [4] M. Adams and J. Ibañez-Guzmán, “Safe Path Planning & Control Constraints for Autonomous Goal”, IROS 2002, Lausanne, Switzerland, 2-5 October 2002. [5] P. Chen, J. Ibañez-Guzmán, T.C. Ng, A.N. Poo and C.W. Chan, “Supervisory Control of an Unmanned Land Vehicle”, Proceedings of 2002 IEEE Symposium on Intelligent Control (ISIC), Vancover, USA, 27-30 October 2002. [6] A. Tay, J. Shen, J. Ibanez-Guzman and C.W. Chan, “Autonomous Vehicle Navigation Strategies - Localised Navigation with a Global Objective”, Proceedings of 2002 International Conference on Information Technology and Application, Australia, 2528 November 2002. [7] G.G. Wang, S.H. Wang and C.W. Chen, “Design of turning control for a tracked vehicle”, Control Systems Magazine, Vol. 10(3), pp. 122-125, (1990). nd [8] J.Y. Wang, Theory of Ground Vehicles (2 Edition), John Wiley, London, (1993). [9] H. Durrant-Whyte, “An Autonomous Guided Vehicle for Cargo Handling Applications”, Int. Journal Robotics Research, Vol. 15(5), pp. 407-440, (1996). [10] A.T. Le, D. Rye and H. Durrant-Whyte, “Estimation of track-soil interactions for autonomous tracked vehicles”, Proceedings of IEEE International Conference on Robotics and Automation (ICRA’97), Albuquerque, NM, Vol. 2, pp. 1388-1393, (1997). [11] S. Scheding, G. Dissanayake, E.M. Nebot and H. Durrant-Whyte, “An Experiment in Autonomous Underground Navigation of an Underground Mining Vehicle”, IEEE Transactions on Robotics and Automation, Vol. 15(1), pp. 85-95, (1999).

(a)

(b) Fig. 7. Control system field trial results. (a) System step responses. (b) Under teleoperation control.

6

INDUSTRIAL SIGNIFICANCE

The successful conversion of a highly mechanical large vehicle into a drive-by-wire system and its deployment as an autonomous vehicle allow SIMTech to demonstrate the capability in autonomous ground vehicle system development. It shows that almost any large moving platform can be converted into a computer controlled one, which could then be deployed as an autonomous or tele-operated vehicle. With the fast advance of autonomous technology in recent years, it’s promising to see that real applications of AGVS will be found in various industrial sectors in a near future. REFERENCES [1] H. Durrant-Whyte, A Critical Review of the State-of-the-Art in Autonomous Land Vehi-

6