Why Machine Vision?

The question "Why do machines really need vision?" could be very easy to answer if we know what part robots play in our society today. As engineers continue to research and improve their (i.e. robots) functionalities, it becomes quite simple to notice why there's need to provide them with some sort if vision (or the ability to see). This advancement in robotics has become an integral part of machines that almost all that operate autonomously have some kind of vision system embedded on them.

As the field of robotics continue to advance on daily basis, engineers will contnue to design and build robots that can behave as close to humans as possible and implementing vision on them is not an exception. Machines with vision, like humans, are able to interact with their environment in a way others cannot. This goes a long way to show why 'machine vision' as a field in robotics engineering can not and should not, be underestimated.

To fully understand why robots need vision, it is paramount that we comprehend why humans need to see and how they use their vision to interact with their environment. Now, imagine yourself without eyes. To do this, close your eyes for a few second and try to navigate your way around the room or a path. You'll notice that no matter how acquainted you are with the path or room, you'll find it extremely difficult to walk around without bumping into objects such as tables, trees, shrubs, walls, chairs, and you name it - depending on where you choose to carry out the excercise. There are other examples that could enlighten us as to how important vision is to us but the above should serve a much simpler purpose.

For autonomous machine (i.e machine that operate without any human intervention), vision is just as important as it is to humans. We want robots that can carry out assigned duties accurately without running into objects within its environment. What better way is there to accomplish this without vision? For example, if we want a simple 'pick and place' robot, we'll design it such that it can see the object(s) it's suppose to pick and also, it must be able to see where to accurately place the object(s). Other robots such as the autonomous vehicles will not properly navigate its environment without the ability to see in some way.

To implement vision on machines, we often use sensors in line with cameras. Some examples of sensors that could be used are photo sensors - sometimes called photodetectors - (e.g line followers, infra-red sensors, and scanning lasers just to name a few) - tells the variation in the amount of light present on a surface or around a visinity. Another example is the motion sensor or motion detector - which tells the presence of an object within its visinity and determines wether to activate the robot or turn it in the direction the object is headed. This could be used in security systems where objects crossing the robot's path are considered as possible intruders. One last example is the proximity sensor. It measures the distance from the robot to any obstacle present within its path.

Cameras, on the other hand, vary depending on the size of the robot and the task it is required to perform. For instance, we might want a robot that can be controlled over the internet using a webpage. To accomplsih this, we need to see what the camera sees and read what the other sensors read on whatever webpage we set up for the control activity. In this example, we want a camera that is capable of taking pictures or videos and sending it over the internet using some sort of controller circuitries and components that are embeded within the robot. The robot should also be capable of recieving instructions from the user webpage at anytime.

From the above dicussions, it should be much easier now to understand why robots need vision; in other words, why they need to see. We don't want machines making mistakes or jamming into other objects around them. We want them to be capable of mimicing the several human behaviours as much as possible without blemish. The field of machine vision in engineering is still relatively new and will only continue to improve as time goes on.

The History of Robots