WO2006087542A1

WO2006087542A1 - Vehicle location

Info

Publication number: WO2006087542A1
Application number: PCT/GB2006/000527
Authority: WO
Inventors: Savan Bhagvanji Chhaniyara; Yahya Hashem Zweiri; Kaspar Alexander Althoefer; Lakmal Dasarath Seneviratne
Original assignee: Kings College London
Priority date: 2005-02-18
Filing date: 2006-02-15
Publication date: 2006-08-24
Also published as: GB0503392D0

Abstract

A method for determining vehicle location, particularly the location of unmanned ground vehicles such as mobile robots, by determining the slip of the vehicle using a camera to take images of the surface as the vehicle moves.

Description

Vehicle Location

The present invention relates to a method for improving vehicle location particularly the location of unmanned ground vehicles such as mobile robots and for accurately determining the speed of vehicles.

To facilitate mobile robot navigation, it is advantageous to provide a mobile robot with information about the precise position and orientation of the mobile robot relative to a fixed coordinate system. Such information may help a mobile robot in planning and following a route, in docking manoeuvres, and in coordinating with other systems, including other mobile robots.

Commonly used methods for providing a robot with such information include dead reckoning, inertial navigation, laser positioning, and satellite-based positioning. Each of these techniques has disadvantages. With dead reckoning and inertial navigation, the autonomous vehicle keeps track of its various movements and calculates its position and orientation accordingly. However, inaccuracies in tracking or calculation are cumulative, so that a series of small errors can lead to substantial mistakes, especially in docking.

Satellite-based positioning requires a clear communications path to a satellite and also does not provide location information with the precision required for docking manoeuvres. Other methods for providing mobile robots with precise position and orientation information involve placing specially designed objects at known positions in the working environment, to be used as landmarks. These landmarks are sensed by a vision system carried by the mobile robot. The visual information about the landmarks is processed by the robot, enabling it to detect and recognize the various landmarks, and to determine its own position with respect to them. US 5999866 describes a method of determining the location of a vehicle by capturing a visual image of the vehicle's present environment and comparing this with a stored map to determine the vehicle location.

Current landmark methods suffer from several disadvantages. Especially if the working environment is cluttered or unevenly lit, or if the landmarks are partially occluded; errors may occur in detecting or recognizing the landmarks, resulting in errors in the position determined for the mobile robot. In addition, the processing in many current landmark systems needed to extract information from visual images requires substantial computational resources, which makes such systems difficult and expensive to implement in applications requiring the mobile robot to determine its position in real time. Landmark systems cannot be used in unfamiliar or unknown territory.

Inertial systems are one of the widely used commercialised totally self-contained navigational systems. Initially developed gyros are of mechanical complex structure and mechanisms involving numbers of moving parts; but rapid development in solid state electronics and Mechatronics has helped to develop electro-mechanical, electro- optical sensors. This type of sensor has no moving parts. This method uses gyroscopes and is combined with accelerometers to measure rate of rotation and acceleration. Measurements are integrated once (or twice) to yield position.

This system is used for missile cruise control, military and civil aviation, space technology for accurate estimations and control. According to these areas of operation, the components used and the entire system has to be very precise and reliable. Consequently, the costs for such a system are still very high and the available cheaper sensors suffer from the accuracy of measurements; moreover the size is not yet as small that it can be used for mobile robots, automotive or consumer electronics. Inertial sensor data drifts with time because of the need to integrate rate data to yield position; any small constant error increases without bound after integration. List count of error is generally in centimetres. This is undesirable for some of the mobile robot application areas like autonomous land mine disabling, radioactive material handling in nuclear plants, scientific research and operations like auto drilling and assembly work in industries.

In motion, autonomous vehicles can suffer from slippage and slippage is one of the big stumbling blocks in achieving accurate localisation of the autonomous vehicles. Behaviour of this phenomenon is very unpredictable, non linear and it changes with the different terrain conditions. This physical property of the slippage has not been taken into account seriously for the control engineering and artificial intelligence point of view and is considered a disturbance. There is a specific need for developing self intelligence in autonomous robots to fulfil this gap and there is the need for a system that scans ground and is able to locate the relative position accurately.

An example of this problem was NASA's Mars Exploration Rover mission which provided an example of how slip plays an important role with the problem of any autonomous vehicle navigation and why there is an immediate need to develop a system to cope with it. The Rover had a hard time gaining traction into the sandy ground around the crater; its wheels spun in place before they actually gained tracking. Mars rover uses an odometer to click off the distance its wheels travels to measure and register how far the vehicle has moved. But they cannot give information about whether a target has or has not actually been reached or how much slippage is occurring.

Apart from its importance in autonomous vehicles, measurement of slippage is important for vehicle safety and operator aids for manned wheeled vehicles and can be used to enhance the driving performance of the vehicle; for example if the slip parameters for each wheel are known, the torque applied to that wheel can be controlled and this can improve braking and prevent wheel spin e.g. in icy conditions.

A method for overcoming this problem is by attaching an extra castor/trailing wheel to the robots' chassis. Generally ball type (spherical) castors or several wheeled castors are attached to the robot with inbuilt encoders on it. This can successfully find the longitudinal slip, lateral slip and slip angle for a high speed robot.

Rotary encoders are attached on axes of individual measuring wheels and castor axes to measure longitudinal and lateral slip. In order to find the longitudinal slip, they use two-sided measuring wheels with encoders. Any difference in the measuring wheel and corresponding driving wheel represents slippage.

This proposed system may be advantageous for the indoor and limited ground specifications but this type of system is totally impractical for the all terrain and unmanned ground vehicles for surveillance, planetary explorations, and military purposes due to the unexpected nature of the terrain and, also, the extra castor wheels limit the mobility of the system and increase complexity.

Methods for determining vehicle slip are described in US 5147010 which determines the slip ratio by measuring the main drive speed and the actual ground speed and calculates the slip and GB 1346678 which determines slip by measuring the vehicle velocity and measuring the rotational velocity of a driven wheel.

We have now devised an improved equipment and system for determining the location of a mobile vehicle which can deal with the problem of slippage.

According to the invention there is provided equipment for determining the slip of a vehicle moving over a surface and in contact with the surface which equipment comprises (i) a means for measuring the actual velocity of the vehicle over the surface, (ii) means for measuring the velocity as determined by the component(s) of the vehicle in contact with the surface and (iii) means for comparing these values.

The invention also provides a method for measuring the slip of a vehicle moving over a surface and in contact with the surface which method comprises measuring the actual velocity of the vehicle over the surface, measuring the velocity as determined by the component(s) of the vehicle in contact with the surface and comparing these values.

The invention also provides a method and equipment for determining the location of a vehicle from the measurement of slip.

The invention also provides equipment for determining the velocity of a vehicle comprising an image forming means able to be attached to the vehicle, which image forming means is able to form an image of the surface on which the vehicle is moving as a spatiotemporal image sequence of the surface by recording a time- varying image sequence and computational means for calculating the velocity of the vehicle from the sequence of spatiotemporal images.

The invention also provides a method for determining the velocity of a vehicle comprising forming a spatiotemporal image sequence of the surface on which the vehicle is traveling by recording a time- varying image sequence and calculating the velocity of the vehicle from the sequence of spatiotemporal images.

The equipment for determining the location of a vehicle preferably comprises an image forming means able to be attached to the vehicle which image forming means is able to form an image of the surface on which the vehicle is moving as a spatiotemporal image sequence of the surface by recording a time-varying image sequence means to process the image sequence to obtain a flow parameter and to calculate the velocity of the vehicle to determine the location of the vehicle.

The method for determining the location of a vehicle preferably comprises forming an image of the surface on which the vehicle is moving as a spatiotemporal image sequence of the surface by recording a time-varying image sequence, processing the image sequence to obtain a flow parameter and calculating the velocity of the vehicle from the flow parameter to determine the location of the vehicle. The term velocity is used in its mathematical sense of measuring the speed and direction i.e. it is a vector not a scalar.

The method enables the slip to be calculated by comparing the velocity of the vehicle as calculated by the direction and speed as calculated from the vehicle propulsion means e.g. driven wheels or tracks etc. with the velocity of the vehicle as calculated by method of the invention.

The equipment can be in the form of a sensor which can be attached to a vehicle. The image sequence is formed by recording a time-varying image sequence from an image recording device such as a fixed or moving camera. A great deal of information can be extracted by recording a time-varying image sequence from the camera. An image sequence (or video) is a series of 2-D images that are sequentially ordered in time. This enables the motion estimation to be calculated.

Selection of the camera depends upon the maximum slide velocity, resolution and compatibility with the image grabber. To capture sufficient overlapping images it requires higher frame rate cameras. Presently available cameras provide 10,000 frames/sec to 10 frames/sec, which are adequate.

By "motion estimation", we mean the estimation of the displacement (or velocity) of image structures from one frame to another in a time sequence of 2-D images.

Any method can be used for extracting motion information from the image sequence and three main techniques are image difference, moving edge detection and motion flow and optical field.

In image difference, basically a moving object is first detected using the difference of images. Then the new position of the object is mathematically calculated with help of pixel moments over time; during the second step of the calculation different aspects are employed like fixed windowing, polygon matching, edge detection etc. In moving edge detection basically the first step is to detect an edge and using -an algorithm to find edges based on colour, shape or brightness patterns or criteria set by the user. Then in the next step, the same edge is traced in the next image at some time interval later; this technique is not usually used standalone and is generally combined with others for better results.

Motion flow and optical field is the preferred method and in this method the motion field assigns a velocity vector to every point in an image. If a point in the environment moves with velocity VQ, then this induces a velocity v_t in the image plane; it is possible to determine mathematically the relationship between vg and v,-

The optical flow field is the velocity field that represents the three-dimensional motion of object points across a two-dimensional image. The motion of each point is described using an optic flow vector. Mathematically, this is the rate of change of the unit vector along the line of sight to the surface point. We can think of this as the motion of the intersection of the line of sight with the view sphere (which moves with the eye or camera).

Features of the optical flow method are that alike features match; a feature in one image must correspond to the matching feature in the other, using some constraints such as geometrical (shape) comparison, comparison of grey levels etc. The assumption is that the appearance of a scene feature does not change significantly between frames.

In this method the image moves consistently; the flow is often locally uniform and discontinuities may occur at scene boundaries. Local flow vectors that are similar reinforce one another. In active vision, the flow will not be uniform close to the fixation point, but will still approximate locally to a simple smooth pattern. The assumption is that the scene is made up of extended reasonably smooth surfaces which move rigidly or at least distort smoothly when moving. The optic flow vectors are small; the amount of motion between frames is small compared to the size of the image and is most likely 10 pixels/frame or less.

There are many different techniques used to determine optical flow and there are many algorithms. In a paper by J. L. Barron, DJ. Fleet and S. S. Beauchemin

"Performance of Optical Flow Techniques" Int. J. Comput.Vis., vol. 12, no 1, pp 43-

77, 1994 and a paper by J. L. Barron, D.J. Fleet, S. S. Beauchemin and T. Burkitt

"Performance of Optical Flow Techniques" Dept. Comput. Sci. Univ Western

Ontario, London, Ontario, Canada, and Dept. Comput Sci. Queens Univ. Kingston, Ontario, Canada, Tech. Rep. TR299, RPL-TR-9107, July 1992, Revised July 1993.

Several of these are described and evaluated.

The software which can be used in the invention is image acquisition software, image processing software and post processing software.

The image acquisition software is generally a type of camera driver. It enables user settings for focus, colour, brightness, shutter speed and acquisition frame settings and enables time based images to be captured and stored.

The image conversion software is used in one of the pre-processing steps. In this step the captured image sequence is manipulated as per the optical flow algorithm criteria. It includes converting images in certain formats like BMP to RAS (Sun Raster Image Format). Some of the filters can be used for sizing or dipping, imaging encasement.

The optical flow algorithm is performed at this stage on the image sequence and results are stored for the post processing stage. The optical flow algorithm consists of sequential reading, performing the processing and then closing the image from memory and displaying the velocity gradients on an image.

Post processing converts the result from the optical flow algorithms into user defined units used for analysis of data. Output velocity of the optical flow algorithms are in — Q —

the pixel/frame units. Depending upon the camera calibration, the results are converted into SI units and this velocity information is then used for detecting the slip parameters.

The invention can provide a system which integrates optical flow measurements with a slip parameter estimation technique, for example, sliding mode observer, Kalman Filter, Extended Kalman Filter, Direct Mathematical Inversion in order to determine the slip parameters; we have found that the sliding mode observer is the preferred method for slip estimation.

From the velocity the positional data can be obtained via integration and it is then possible to plot the present trajectories and the occurring slip. This data is then fed back to a trajectory controller for correction of the trajectory controller to enable the vehicle to follow a tight trajectory.

The invention is particularly useful with unmanned ground vehicles (UGV) and mobile robots and enables their location to be determined from measuring the velocity of the vehicle or robot from its starting point.

As set out below the invention can predict side slip and by comparing the velocity of the vehicle or robot from the driving means in contact with the surface over which it is moving, and comparing this with the velocity calculated by the method of the invention it is possible to calculate the slippage.

As well as being used in unmanned vehicles, the invention can be used with driven vehicles such as commercial vehicles to provide for increased safety and improved navigability. The invention enables slip in x and y direction as well as the angular component of slip to be determined. This was not possible in earlier anti-skid systems. Computing the difference between speed measured at the wheels and the vehicle speed (against ground) as done by the method and equipment of the present invention, these three slip components can be accurately determined. This aspect is very important for any advanced traction control system in a vehicle. With the slip components accurately computed, the actuation signal for each wheel (i.e. power requirements for each wheel) as well as the steering angle can be calculated to achieve the most appropriate vehicle trajectory for a given situation (i.e. to drive the vehicle around a bend if one or more wheel is suddenly experiencing high slip due to changes in road surface (e.g. icy or wet roads).

Such an advanced system of vehicle control would take over control from the driver and attempt to steer the vehicle out of a dangerous situation as safely as possible.

As wejl as its use for measuring location and velocities of vehicles, the present invention could provide estimates on location when a GPS system fails to provide information on position (e.g. in a tunnel, in narrow city streets). First of all, the proposed system can provide better results when small accelerations occur and/or when the vehicle is moving at a constant speed (in such cases accelerometer signals are very noisy and unreliable). Secondly, error accumulation is not so high, since

^• only one integration from the speed measurements (output of the system of the present invention) is needed in contrast to two integrations which are needed if accelerometers (inertial sensors) are used.

An explanation of how the system of the present invention is set out below.

1. Basic slip models

When driving torque is applied to a wheel, a tractive force is developed at the contact area between the wheel and the ground. During this procedure the wheel's tread in front and within the contact area is under compression forces. These also develop the shear deformation of the sidewalls of the wheel. The distance travelled under action of the driving torque will be generally less then in free rolling. The main reason behind this is that the tread elements are compressed before entering the contact region. This phenomenon is referred to as a longitudinal slip. The linear tyre model doesn't consider lateral tyre forces due to complex interactions between lateral and longitudinal tyre forces.

Mathematically slip can be described as: rθ-x

where rθ Wheel angular velocity

X Vehicle velocity a Slip angle

S Slip

V Side Slip velocity

2. Combined slip model

The combined slip is most likely to come into the picture when the wheel is cornering when side forces also come into effect. This develops a lateral force at a contact area between the wheel and ground. Due to this the wheel will move along a path at an angle α with the wheel plane. This angle α is usually referred to as the slip angle. Combined slip demonstrates a strong nonlinear behaviour; it depends on a number of the parameters like soil type, tyre soil adhesion, side forces, pressure on the wheel and many more. It is very complicated to derive an accurate mathematical model of the combined slip but basically we can put it as below.

S_v = tm a 0.2)

a = tan — x ^r 0 Wheel angular velocity x Vehicle velocity a Slip angle

S Slip

Side Slip velocity 3. Prediction of slip

The prediction of the slip requires two important parameters - angular velocity and vehicular velocity, either for linear wheeled robots, or for tracked vehicles. Angular velocity can be obtained from the encoders mounted on driving sprockets.

Velocity of the UGV is obtained by this proposed system and using equation 1.1 and 1.2 or it can be incorporated with slip model based on kinematics and dynamics of the tracked vehicle as set out in calculations (1.3-1.4).

y = ^[<°o (¹ - k ) + ^ωι (¹ - 0] [si n φ{t) - ∞sψ(ή tan a(ή] (1.3)

, Δv _ φv_D(1--_if₀)->v,(1-/,)] ^Ψ B B

x

B -

cu = arctan(— ) -φ x

Where

S Slip for tire model r Radius of wheel rθ Angular velocity of driving sprocket

V Side slip velocity

Xy Vehicle velocity θ_\ Front caster angle orientation θ₂ Back caster angle orientation

A contact area, area i slip i_j slip of the inside track of a tracked vehicle i₀ slip of the outside track of a tracked vehicle

IC shear deformation modulus b width of a single track

I length of a tracked vehicle

Iz mass moment of inertia of vehicle about the Z axis

C cohesion φ angle of internal shearing resistance

B the tread of a tracked vehicle, i.e. the spacing between the centreline of the two tracks

F total tractive effort of a track φ angular displacement of a tracked vehicle

V₀ speed of outside track of a tracked vehicle

V_I speed of inside track of a tracked vehicle

V_x the velocity component of vehicle along the X _e axis

V_c the velocity of vehicle

X linear displacement of a tracked vehicle along X axis with regard to arbitrary frame y linear displacement of a tracked vehicle along Y axis with regard to arbitrary frame α slip angle μ_r coefficient of traction μ._| coefficient of lateral resistance

R turning radius

O instantaneous centre of rotation

W weight of tracked vehicles

Fo.i thrust on the outer, inner track, Λ'

X₁, linear displacement of a tracked vehicle along X axis with regard to frame attached to the vehicle. y_e linear displacement of attacked vehicle along Y axis with regard to frame attached to the vehicle φ angular displacement of a tracked vehicle about Z axis, measured clock O)₀ the angular velocity of outer track

(Oj the angular velocity of inner track

Thus this system measures slip independently and it can also predict the slip using the above model. It is a feature of the invention that the system is totally self contained; accuracy is very high as the resolution is limited only by optics; capital cost is limited to that of a camera, lighting, offline storage and a processor; installation cost is limited to the labour and time required for initial setting up and it is a totally non contact type sensor/method for measuring the slip and it does not restrict the mobility of the UGV in any type of terrain.

The invention can be used with a trajectory controller which controls the trajectory of a remote vehicle by controlling the velocity of the vehicle. In this application information about the location of the vehicle is fed to the trajectory controller, compared with the desired trajectory and the velocity of the vehicle adjusted by the trajectory controller to correct any deviation from the desired trajectory.

This gives a superior result compared with trajectory controllers which rely on the speed of rotation of wheels or tracks and measurements of distance travelled which, as explained above, are subject to errors due to slippage.

An illustration of a vehicle incorporating the present invention is shown in fig. I₅ in which :- Fig. 1 shows a tracked vehicle and

Fig. 2 shows a basic flow diagram of the system

Referring to fig. 1 a tracked UGV (1) moving over terrain (2) has a camera (3) mounted on it. The camera (3) takes pictures of the terrain perpendicularly below it at (4). As the vehicle moves over the terrain the images obtained by the camera are processed according to the invention.

Referring to fig. 2 the images from camera (3) are processed and a spatiotemporal image sequence obtained. This is processed by the optical flow algorithm to obtain the flow parameter which gives the velocity and position, this parameter passed to the central control algorithm along with the reading from the encoder on the UGV which gives the readings from the drive chain of the UGV. The central control algorithm calculates the slip and a correction based on the kinematics and dynamic model of the UGV can be applied. This can be passed onto the trajectory controller to give a corrected trajectory.

In the Example a system on test scale was evaluated.

Example

One dimensional motion was investigated for experimental purposes. A camera is mounted onto a carriage which can move only in one direction and the carriage represents the moving unmanned robot. Different terrains were represented by sand, small stony terrain, concrete and carpet.

Stage 1 Camera Calibration The camera was calibrated to establish the relationship between the image coverage of the terrain in x and y directions with the image resolution dimension in the pixels. Thus this step provides the relationship between image pixels and the trajectory parameters at the mm. scale.

Stage 2 Image Acquisition

Spatiotemporal image sequences were taken by the camera and stored on a hard disk; the hardware used was drom 3Com Home Connect Pc Webcam and the software was Capture Studio Professional 4.1.

The carriage was moved in one direction with a uniform motion and images were taken at a predefined rate; the maximum frame rate of the camera was 15 frames/sec. The motion of the camera was measured by a fixed rule and the distance moved during each frame registered manually. The BMP Image format was used for image capture. Stage 3 Image Conversion

The image sequence was converted into a desirable format before processing; for the image conversion and enhancement ACD See 4.0 software was used. The acquired image sequence was in BMP form and was converted to Sun Raster Image Format (RAS).

Stage 4 Image Processing Optical Flow Analysis

After image acquisition, an optical flow algorithm is used to find the velocity vectors in the direction of motion of the carriage. The implementation of the Phase based approach to the estimation of the optical flow field using spatial filtering on MATLAB is used. This software is freely available on MATLAB Central File exchange. This algorithm proceeds in 3 steps:

1. Spatial filtering 2. Phase gradient estimation

3. IOC using recurrent networks

The flow chart of the optical flow algorithm is shown in Table 1.

Table 1

The output of the algorithm consists of the velocity of each moving pixel in x and y direction.

Stage 5 Post Processing In this step an average velocity component is calculated from the results available from Stage 3 above and it is converted into trajectory, parameters. For establishing the trajectory of UGV, instantaneous velocity data were plotted against time and then a curve filling model established. After integrating the curve fitting model we can get the position at different time intervals. We can process this data for further analysis or it can be used as a feedback for a robot trajectory controller.

Results

Experiments were performed for the manually archived image sequences for different terrains and the results correlated with the actual results.

Results of the Experiments are shown below.

Parameters Terrain Sample: Sand Frame Rate : 30 frames/sec Actual Distance moved: 100mm

.Experiment: 1

Avg velocity 2.428069337

Actual Distance Moved 100 mm

Total distance 249.7101509_pixels

95.20199503 mm relative error 4.798004966

Experiment 2

Total Dist 244.2304 pixels Total Dist 93.1128 mm Relative Error 6.89%

Total Dist 276,6976 pixels Total Dte\ l05A910mm Relative Error -5.49 % Summar of Results

The results are shown graphically in figures 3 to 5 of the drawings.

As can be seen the results obtained by the method of the invention correlate accurately with the actual movement.

Claims

1. Equipment for determining the slip of a vehicle moving over a surface and in contact with the surface which equipment comprises (i) a means for measuring the actual velocity of the vehicle over the surface comprising an image forming means able to be attached to the vehicle, which image forming means is able to form an image of the surface on which the vehicle is moving as a spatiotemporal image sequence of the surface by recording a time-varying image sequence; (ii) processing means able to process the image sequence to obtain a flow parameter means to measure the actual velocity of the vehicle; (iii) means for measuring a velocity of the vehicle as determined from component(s) of the vehicle in contact with the surface; (iv) means to calculate the slip of the vehicle from the flow parameter and the velocity as measured by the component(s) of the vehicle in contact with the surface and (v) means for comparing these values.

2. Equipment as claimed in claim 1 in which the image forming means comprises a camera.

3. Equipment as claimed in claim 2 in which the camera provides at least 10 frames/sec.

4. Equipment as claimed in any one of claims 1 to 3 in which the processing means comprises a processor which is able to process the image sequence by a motion flow and optical field algorithm.

5. Equipment as claimed in claim 3 or 4 in which the processing means is able to display the velocity gradients of the vehicle on an image.

6. Equipment as claimed in any one of claims 1 to 5 in which the spatiotemporal image sequence is processed by a process selected from image difference, moving edge detection and motion flow and optical field.

7. Equipment as claimed in any one of claims 1 to 5 in which the image of the surface is processed by image acquisition software, image processing software and post processing software.

8. Equipment as claimed in claim 7 in which the image acquisition software is a camera driver which enables user settings for focus, colour, brightness, shutter speed and acquisition frame settings and enables time based images to be captured and stored.

9. Equipment as claimed in claims 7 or 8 in which the image conversion software manipulates the captured image sequence as per the optical flow algorithm criteria and converts images to a different format.

10. Equipment as claimed in claim 9 in which the image is converted from BMP to RAS (Sun Raster Image Format).

11. Equipment as claimed in claims 9 or 10 in which the optical flow algorithm consists of sequential reading, performing the processing and then closing the image from memory and displaying the velocity gradients on an image.

12. Equipment as claimed in any one of the preceding claims in which the means to measure the velocity of the vehicle as determined by the component(s) of the vehicle in contact with the surface are encoders connected to the drive train of the vehicle.

13. Equipment as claimed in any one of claims 1 to 12 in which is a sensor able to be attached to a vehicle.

14. Equipment as claimed in any one of claims 1 to 13 in which the vehicle is an unmanned ground vehicle or a robot.

15. Equipment for determining the velocity of a vehicle comprising an image forming means able to be attached to the vehicle, which image forming means is able to form an image of the surface on which the vehicle is moving as a spatiotemporal image sequence of the surface by recording a time-varying image sequence and computational means for calculating the velocity of the vehicle from the sequence of spatiotemporal images.

16, A method for measuring the slip of a vehicle moving over a surface and in contact with the surface which method comprises measuring the actual velocity of the vehicle over the surface by forming an image of the surface on which the vehicle is moving as a spatiotemporal image sequence of the surface by recording a time-varying image sequence, measuring the velocity as determined by the component(s) of the vehicle in contact with the surface and comparing these values.

17. A method as claimed in claim 16 in which the actual velocity of the vehicle is measured by forming an image of the surface on which the vehicle is moving as a spatiotemporal image sequence of the surface by recording a time-varying image sequence, processing the image sequence to obtain a flow parameter and calculating the velocity of the vehicle from the flow parameter.

18. A method as claimed in claim 17 in which the image of the surface is formed by a fixed or moving camera.

19. A method as claimed in claim 18 in which the camera provides at least 10 frames/sec.

20. A method as claimed in any one of claims 15 to 18 in which the spatiotemporal image sequence is processed by a process selected from image difference, moving edge detection and motion flow and optical field.

21. A method as claimed in any one of claims 16 to 20 in which the image of the surface is processed by image acquisition software, image processing software and post processing software.

22. A method as claimed in claim 21 in which the image acquisition software is a camera driver which enables user settings for focus, colour, brightness, shutter speed and acquisition frame settings and enables time based images to be captured and stored.

23. A method as claimed in claim 21 or 22 in which the image processing software manipulates the captured image sequence as per the optical flow algorithm criteria and converts images to a different format.

24. A method as claimed in claim 23 in which the image is converted from BMP to RAS (Sun Raster Image Format).

25. A method as claimed in claims 23 or 24 in which the optical flow algorithm consists of sequential reading, performing the processing and then closing the image from memory and displaying the velocity gradients on an image.

26. A method as claimed in any one of claims 16 to 25 in which the velocity as determined by the component(s) of the vehicle in contact with the surface is obtained by encoders connected to the drive train of the vehicle.

27. A method for determining the velocity of a vehicle which comprises forming an image of the surface on which the vehicle is moving as a spatiotemporal image sequence of the surface by recording a time- varying image sequence, processing the image sequence to obtain a flow parameter and calculating the velocity of the vehicle from the flow parameter..

28. A method as claimed in claim 27 in which the image of the surface is formed by a fixed or moving camera.

29. A method as claimed in claims 27 or 28 in which the camera provides at least 10 frames/sec.

30. A method as claimed in any one of claims 27 to 29 in which the spatiotemporal image sequence is processed by a process selected from image difference; moving edge detection and motion flow and optical field.

31. A method as claimed in any one of claims 27 to 30 in which the image of the surface is processed by image acquisition software, image processing software and post processing software.

32. A method as claimed in claim 31 in which the image acquisition software is a camera driver which enables user settings for focus, colour, brightness, shutter speed and acquisition frame settings and enables time based images to be captured and stored.

33. A method as claimed in claim 31 or 32 in which the image processing software manipulates the captured image sequence as per the optical flow algorithm criteria and converts images to a different format.

34. A method as claimed in claim 33 in which the image is converted from BMP to RAS (Sun Raster Image Format).

35. A method as claimed in any one of claims 33 or 34 in which the optical flow algorithm consists of sequential reading, performing the processing and then closing the image from memory and displaying the velocity gradients on an image.

36. A method of controlling the trajectory of a vehicle which comprises determining the location of the vehicle by the method of any one of claims 27 to 35 feeding the information about its location to a trajectory controller to enable the trajectory controller to adjust the velocity of the vehicle to the required trajectory.

37. A method for determining the location of a vehicle comprising measuring the velocity of the vehicle as claimed in any one of claims 27-36 and using the velocity to determine the location of the vehicle.