Are you interested in our solutions? Please don't hesitate to contact us! We will be glad to help you.Start enquiry
Artificial intelligence (AI) is the term used to describe a process by which machines learn how to learn. Computer systems mimic human intelligence by performing intelligent behavior based on predefined or learned models. An AI system perceives its surroundings through e.g. cameras and sensors; it then interprets this information and derives actions from it.
Machine learning and deep learning can be classified under the generic term artificial intelligence. Machine learning describes a process by which computers can acquire knowledge autonomously. The most common method for this now is deep learning. With deep learning, a computer acquires its knowledge through an algorithm that analyzes large amounts of data and learns to draw conclusions from this data. For example: We want the algorithm of a system to learn how to identify stop signs. So, we show it one million images of stop signs, and the algorithm builds up a store of experience. Based on this, it is then able to identify other stop signs presented to it.
The core intelligence of automated vehicles lies in the software running on the vehicle computer in the form of adaptive algorithms. The software analyzes and interprets incoming data from the surroundings sensors. It can, for instance, identify whether an object perceived by the sensors is a stop sign, vehicle, pedestrian, or cyclist. It can furthermore determine whether the object is moving and, if so, in which direction and at what speed. Based on the interpretation models, it is also possible to derive the likely future behavior of these static or dynamic objects. This is how it learns the highly advanced skill of anticipation: After observing a very large number of objects during the development process, the system is able to determine their respective characteristic behavior, which helps it to make increasingly reliable predictions. This is similar to how humans can learn from experience. In future, thanks to the vehicle’s intelligence, it will be possible, for instance, to calculate the probability of a pedestrian crossing the path of the vehicle and so enable the brake system to be activated in good time. Its swift powers of perception and extremely fast response times mean that AI is a very powerful technology that will find its way into assisted and automated vehicles and, unlike a driver, will not tire.
The human being must be enabled to act as the controlling authority in all AI decisions. “Artificial intelligence is meant to serve humans. With the AI Code, we have provided our employees with clear guidelines for the development of smart products,” said Bosch CEO Volkmar Denner at the IoT industry gathering Bosch ConnectedWorld (BCW) in Berlin. “We want people to trust our AI products.”
Automated vehicles must have three fundamental skills: They need to perceive and interpret (Sense) their surroundings, use this information to make forecasts and derive a suitable driving strategy (Think), and then implement it reliably and safely (Act). Sensing the immediate environment is the task of the surround sensors which combines camera, radar and ultrasonic. Vehicle intelligence enabling it to interpret its surroundings and find the optimal driving strategy is made possible by software and algorithms which use the information gleaned from sensors as well as data from other connected systems.
Systems such as the powertrain, steering and brakes ensure that the respective driving strategy is then implemented on the road. This process of sensing, thinking and acting takes place during the entire journey. When applied to human beings, this process is similar to constant interactions in the body during which the sensory organs pick up stimuli which are processed by the brain before the nerve pathways, muscles and limbs implement the brain's control signals as actions.