Self-driving Cars

What are Self driving Cars?

A Self-driving Car, also known as an autonomous vehicle (AV), is a vehicle that can perceive the surrounding environment and navigate with little or no human intervention. Self-driving cars combine many heterogeneous sensors, actuators, high-end processors, complex algorithms for perception, localization, planning and control. To perceive the surrounding environment, they include many sensors such as lidar, GPS, and many more. Information from these sensors is interpreted by advanced control systems to identify navigation paths, detect any obstacles and identify relevant signages. In simple words, a self-driving car is a computer-controlled car that drives itself.

self driving cars

Classifications of self driving cars:

An automotive standardization body has classified self-driving cars into a six-level classification system ranging from fully manual to fully automated.    

  • Level 0: In this level, the automated system issues warnings and has no sustained motor control. There is no automation; full-time performance by a human driver is needed under all circumstances.
  • Level 1: They are also known as hands-on or driver assistance mode. In this, both the driver and the automated system share the vehicle control, such as parking assistance, cruise control, lane-keeping assistance and collision mitigation system.
  • Level 2: This level is also known as hands-off or partial automation mode. The automation system takes full control of the vehicle including accelerating, steering, and braking. However, the driver must monitor the driving and if necessary, intervene to take complete control of the vehicle.
  • Level 3: This level has a conditional automation system and is known as eyes off mode. There is a necessity for drivers but can safely turn their attention away from driving tasks. 
  • Level 4: This level has a high automation system in place, and there is no requirement of driver attention. Therefore, a driver may safely leave the driver’s seat or go to sleep. Hence, this level is also termed as mind off mode. 
  • Level 5: No human intervention is required, and even the steering wheel is optional. A full automation system is in place to provide a complete experience of an autonomous vehicle.

Architecture of self driving cars:

The architecture of self-driving cars is typically organized into the perception system and the decision-making system with many sub-systems among them. The perception system is responsible for tasks such as localization, obstacle and road mapping, among others. The decision-making system is responsible for tasks like route planning, path planning, motion planning, obstacle avoidance and control.  

  • Perception system: As the name suggests, they are responsible for creating a representation of the surrounding environment, using onboard sensors, road network, traffic rules, car dynamics, etc. The system further divides into many sub-systems, and few are listed below.   
    • Sensors: They are the main component of a self-driving car. Many sensors are placed in an autonomous car to capture the necessary information of the surrounding environment. A few important sensors are Light Detection and Ranging (LIDAR), Radio Detection and Ranging (RADAR), camera, Global Positioning System (GPS), Inertial Measurement Unit (IMU), odometer, and many more.    
    • Localizer: It is responsible for estimating the state of the car such as pose, linear velocities, angular velocities, and others. 
    • Mapper: This sub-system is fed with localiser data and offline maps to generate the online map. These online maps are used by a certain sub-system of decision making.   
    • Moving object tracker (MOT): It uses the data from localizer and offline maps, to calculate the pose and velocity of the nearest moving obstacles such as other vehicles and pedestrians. 
    • Traffic signalization detector (TSD): This sub-system is responsible for the detection and recognition of traffic signalization. It uses sensors’ data and localizer information.  
  • Decision-making system: It is responsible for navigating the vehicle from the initial position to the final destination defined by the user, considering the internal representation of the environment and the current car’s state, in addition to traffic rules and passengers’ safety and comfort. This system is also divided into main subsystems as noted below.
    • Route planner: This sub-system computes a route, considering the current state of the car and user given final destination. A route is a sequence of waypoints, where each waypoint is a coordinate in the offline maps.
    • Path planner: It computes a set of paths, considering the information from the localizer, traffic rules, and internal representation of the environment. A path is a sequence of poses, where each pose is a pair of coordinates in the offline maps.  
    • Behaviour selector: It is responsible for choosing the current driving behaviour based on the data received from the path planner. This sub-system also selects the pose and the velocity of the car, considering the current driving behaviour and avoiding collision within the decision horizon time frame.
    • Motion planner: This is responsible for computing a trajectory, considering the data from route planner, behaviour selector, satisfies the car’s kinematic and dynamic constraints, and provides comfort to the passengers.   
    • Obstacle avoider: This sub-system receives the information from the motion planner. It analyzes the trajectory for obstacle avoidance and if necessary, changes the trajectory chosen by motion planner to overcome the obstacles. 
    • Controller: It takes the data from obstacle avoider sub-system, and computes and sends the required commands to the actuators of the steering wheel, throttle, and brakes to complete the trajectory, and reach the final destination without any discomfort.

Applications

This technology is mainly used for personalized mobility, public transport system and goods carrying trucks. Many automobile manufacturers around the world are developing and testing autonomous vehicles. Some companies have already launched their semi-autonomous and self-driving cars. A few prominent self-driving cars are listed below.

  • Google launched its self-driving project in 2009, and today it is named as Waymo LLC. Today, they operate and offer commercial self-driving cars that are fully autonomous. 
  • Tesla offers self-driving car capability among all its models. 
  • General motors have the second largest autonomous vehicles under testing as of now. 
  • Argo AI from Ford company is also testing its self-driving cars.
  • Baidu, a technology giant in China, is also testing its autonomous vehicles on Chinese roads.  
  • Otto and Starsky robotics are focusing on autonomous trucks.
  • Online grocers and Ocado are concentrating on autonomous vans.