Self-Driving Cars Navigate

Self-driving cars could prevent the deaths and injuries caused by human error in traffic. This is your VocaCast briefing on autonomous vehicles for Wednesday, April 22.

First, the urgency driving the technology — then the scale of the opportunity ahead.

Human error is cited as the cause for over 90 percent of traffic accidents, which explains the imperative behind autonomous navigation. ^[1] When machines handle the wheel, they don't get distracted, don't fall asleep, and don't make split-second judgment calls that kill. Autonomous vehicles are anticipated to significantly reduce traffic congestion and accidents, alongside improving passenger safety and fuel consumption.

The economic promise is enormous. According to a report by McKinsey, the global autonomous vehicle market is projected to reach seven trillion dollars by 2050, with the number of self-driving vehicles on the road expected to reach 55 million by 2030. ^[2] The market for Advanced Driver-Assistance Systems and Autonomous Vehicles is projected to generate 400 billion dollars in revenue by 2035. ^[2] Autonomous vehicles offer potential for new mobility services, such as robotaxis and autonomous trucking.

Accessibility matters too. Autonomous vehicles can enhance accessibility for diverse populations, including low-income households and persons with disabilities. ^[3] The technology doesn't just promise efficiency — it reimagines who gets to move through the world.

Building that vision of autonomous vehicles requires machines that can actually perceive the world. Self-driving cars function as rolling robots that continuously sense their surroundings, choose actions, and implement them to navigate various environments. ^[4] But raw sensor data alone solves nothing. The ability of a vehicle to understand its environment, determine its position, and navigate safely relies on foundational pillars like perception and mapping.

The technologies powering that perception work in concert. Core technologies for autonomous vehicle mapping include LiDAR for three-dimensional representations, cameras for visual data, radar for object and speed detection especially in adverse weather, and GPS and inertial measurement units for location and movement tracking. ^[1] Autonomous vehicles rely on sensors like LiDAR, cameras, and radar to perceive their environment, identifying other vehicles, pedestrians, traffic lights, and obstacles. ^[5] These individual sensors don't operate in isolation.

That fusion process solves one challenge but creates another: knowing where you are while building a map of unfamiliar terrain. Simultaneous Localization and Mapping, or SLAM, is a core technology enabling autonomous vehicles to create a map of their surroundings while simultaneously determining their own position within that map. ^[5] SLAM becomes essential for navigating complex, dynamic, or poorly covered GPS environments, allowing vehicles to adapt to changes like road construction. ^[5] The algorithms work by processing incoming sensor data for perception, estimating vehicle motion, and correcting accumulated errors by recognizing when a location has been revisited.

Beyond real-time sensing, autonomous vehicles also rely on high-definition maps combined with AI and geospatial data to create detailed maps for precise navigation, offering centimeter-level accuracy in lane markings, traffic signs, and curb heights. ^[5]

Mapping the environment depends on layered sensing systems working in concert. Self-driving vehicles use three primary sensor types, each providing critical information that alone would be incomplete.

LiDAR sensors generate three-dimensional point clouds by continuously emitting laser pulses to measure distance and speed. ^[2] This creates detailed 3D maps of the environment in real-time, enabling vehicles to detect objects and measure distances with high precision. ^[5] Cameras add another dimension by providing detailed visual information used for identifying landmarks, lane markings, and traffic signs. ^[5] They process this visual data using computer vision algorithms to identify objects, detect lane markings, and read traffic signs. ^[2] Stereo cameras estimate distance via image disparity, layering depth perception onto the visual feed. Radar systems use radio waves to detect objects and measure their distance and speed, offering resilience to adverse weather conditions where cameras and LiDAR can fail.

Radar is also useful for features like adaptive cruise control. ^[2] This redundancy matters because no single sensor sees everything—rain obscures optics, dense fog defeats lasers. Together, these three systems provide the situational awareness necessary for accurate path planning and safe maneuvering.

Those sensors give the car its eyes, but it also needs to know exactly where it is and how it's moving. That's where positioning and motion systems come in. GPS determines the vehicle's exact coordinates and speed, which are essential for path planning and navigation. ^[6] The system works through a surprisingly elegant geometry problem. GPS satellites transmit radio signals containing their position and the exact time of transmission, traveling at the speed of light. ^[6] A GPS receiver calculates distances to those satellites based on the time it takes for their signals to arrive, then uses trilateration to determine its position. ^[6] The receiver needs a minimum of four satellites to pinpoint location in three-dimensional space—latitude, longitude, and altitude.

But trilateration is just one navigational technique in the toolkit. Triangulation, a separate method, involves determining a location by measuring angles to it from two or more known points. ^[7] Navigation mechanisms overall include triangulation, trilateration, and inertial guidance systems working in concert. ^[8] Beyond GPS itself, a broader constellation called GNSS, or Global Navigation Satellite System, encompasses systems like GPS, GLONASS, BeiDou, and Galileo that provide global navigation and positioning coverage. ^[6] The core components of GNSS include satellite constellations, ground infrastructure, receivers, and various positioning techniques such as Differential GNSS and Real-Time Kinematics.