LiDAR has become a popular buzzword that describes expensive, high performance sensors used on self-driving cars. The three dimensional images seen in magazines and on websites tell a singular story of science fiction technology brought to life at great expense. Is this the whole story or is there more? LightWare believes that we have just scratched the surface of LiDAR’s potential and in order to realise this potential we have to look in a different direction, through the eyes of machines.
Many machines need to be aware of the objects and spaces around them. This might be to detect when something arrives or leaves, to measure how high something is or to see how far away an obstacle is. It is the position of things relative to the machine that affects the decisions that the machine must make. Human centric visual characteristics are not important to a machine. It is the pure logical concept of presence, absence and position that determine what actions must be taken. This is where LiDAR excels as it clearly identifies objects and can measure the spaces between them.
A machine actually perceives the world through data, not eyes and ideally this data should contain only critical and relevant information. Any unnecessary complexity would need analysing and filtering at the expense of processing overhead. Machine Learning, AI and quantum computing have been invoked as solutions to analysing complex data but these may make a machine too expensive or complicated to build. A more effective solution is to provide the right data in the first place and this is where LiDAR is so attractive. It provides information about the location of objects that is easy for a machine to understand.
So while we like to give our favourite machines anthropomorphic character, the truth is that a machine doesn’t need to watch a thousand cat videos in order to step over a cat if it has LiDAR perception. The LiDAR shows the cat as a cloud of data points that are clearly an obstacle, even if the machine has never seen a cat before. Just as importantly, a machine with LiDAR perception does not need to have numerous cameras recording every detail of your life. Instead, it senses where things are located without taking any pictures.
So why aren’t machines already using LiDAR perception? Price. Up until now commercially available LiDAR systems have been too expensive to add to an autonomous lawn mower or a delivery drone. LightWare has removed this final roadblock by offering a family of LiDAR sensors specifically tailored for machine perception at a very affordable price.