Google’s self-driving car is now the Waymo Driver: a continuously operating system that combines detailed maps with lidar, cameras, radar, audio, onboard computing, machine-learning software, and vehicle controls. It determines where the vehicle is, interprets the live scene, predicts what nearby road users may do, and plans and executes a safe maneuver within its operating conditions.
Google’s self-driving car is not powered by one camera, one map, or one remotely controlled computer. The technology began as Google’s Self-Driving Car Project in 2009 and is now developed and operated by Waymo as the Waymo Driver. It combines detailed maps, lidar, cameras, radar, audio sensors, onboard computing, machine-learning software, vehicle controls, simulation, real-world testing, and operational support.
As an Amazon Associate I earn from qualifying purchases.
In every driving cycle, the system repeatedly answers four questions: Where am I? What is around me? What will happen next? and What should I do? It matches live sensor data to mapped road features, builds a model of nearby traffic and pedestrians, predicts possible movements, selects a safe route and trajectory, and converts that decision into steering, braking, and acceleration commands.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →From Google’s project to the Waymo Driver
Google began its self-driving-car work in 2009. Early testing used modified Toyota Prius vehicles, followed by purpose-built prototypes such as the Firefly. The project eventually progressed from experimental testing to fully autonomous public-road rides, including an early fully autonomous ride in Austin in 2015.
#1 Best Overall
The project later became Waymo, an Alphabet company focused on autonomous-driving technology and autonomous ride-hailing. That history matters because today’s system is not simply the original Google prototype with newer software. Waymo has developed multiple generations of sensors, computers, vehicle integrations, maps, software, simulation tools, and operating procedures.
It is more accurate to think of the Waymo Driver as a portable autonomous-driving system than as one particular car. The Driver is integrated with different vehicle platforms, while the underlying technology includes the sensing, computing, mapping, prediction, planning, control, and safety-assurance systems that make autonomous driving possible.
The four questions behind autonomous driving
| Question | What the system must determine |
|---|---|
| Where am I? | The vehicle’s precise position and orientation relative to the road, lanes, traffic signals, curbs, and other mapped features. |
| What is around me? | The locations, types, movements, and relevant properties of vehicles, pedestrians, cyclists, signs, signals, construction zones, road edges, and other objects. |
| What will happen next? | The plausible future movements of nearby road users, such as a pedestrian approaching a crosswalk or a vehicle preparing to merge. |
| What should I do? | The route, lane choice, speed, yielding behavior, steering path, and other actions that allow the vehicle to progress safely. |
These are not four steps that run once and then finish. They form a continuous feedback loop. New sensor observations update the vehicle’s location and world model; those updates change predictions; and the plan is revised as traffic, pedestrians, weather, and road conditions change.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstall1. Detailed maps and localization: knowing where the car is
Waymo prepares detailed maps of the areas where its vehicles operate. Those maps can include lane markings, stop signs, traffic signals, curbs, crosswalks, lane geometry, and other relatively stable road features.
During a ride, the vehicle compares its live observations with those maps. This process, called localization, gives the Driver a precise local reference. GPS can contribute positioning information, but GPS alone is not sufficient for autonomous driving—signals can be degraded or unavailable in some environments, and GPS does not describe the exact position of the vehicle within a lane or explain every road feature around it.
A map is not a frozen description of everything the vehicle will encounter. It provides a structured reference for stable features. Sensors provide the changing information: a pedestrian entering a crosswalk, a cyclist moving around a parked car, a temporary stop sign, road debris, construction, a newly blocked lane, or another driver’s unexpected maneuver.
This division of labor explains why autonomous driving is not simply a matter of giving a car a nationwide road database. Before a system can operate reliably in a new area, it needs to understand local road geometry, signage, traffic behavior, weather, construction patterns, regulations, and operating constraints. The mapping and validation work must then be connected to the vehicle’s ability to handle conditions that differ from the map.
Free tools Windows power users keep installed
One-click scans. No signup required.
2. Sensors: how the Driver sees and hears the road
The Waymo Driver uses several sensor types because each one has different strengths and failure modes. The goal is not merely to collect more data. The goal is to combine overlapping information so the system is not forced to rely on one modality in every situation.
Lidar builds a three-dimensional view
Lidar—short for light detection and ranging—sends laser pulses into the environment and measures how long their reflections take to return. Those measurements create a three-dimensional point cloud representing the position and shape of objects around the vehicle.
Lidar is particularly useful for measuring distance and geometry. It can help the system identify the location and outline of vehicles, pedestrians, cyclists, curbs, barriers, and other objects, including objects that are partly hidden from view. Because lidar supplies its own laser illumination, it can provide ranging information in both daylight and darkness.
Cameras provide visual meaning
Cameras supply visual information that complements lidar’s distance measurements. They help the system interpret traffic-light colors, road signs, lane markings, construction features, road surfaces, and other visual details.
Waymo’s descriptions of newer hardware generations include higher-resolution cameras, expanded dynamic range, low-light capabilities, and cleaning systems intended to preserve visibility when the sensor surfaces become dirty or affected by road conditions. A camera is therefore an important part of the system, but it is not the whole system.
Radar measures distance and relative speed
Radar contributes information about an object’s distance and relative speed. It can remain useful in rain, fog, snow, and other conditions in which camera visibility may be reduced.
Waymo’s sixth-generation hardware descriptions refer to imaging radar that provides dense information about object distance, velocity, and size over time. That temporal information can help the system understand how nearby traffic is moving, not merely where it appeared in one camera frame.
External audio receivers add another cue
Waymo’s newer hardware descriptions also discuss external audio receivers. These can help detect and localize sounds such as emergency-vehicle sirens and railroad crossings. Audio is a supplementary input rather than a replacement for cameras, lidar, or radar, but it can provide useful context that visual and ranging sensors may not supply as directly.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Sensor fusion creates a more resilient picture
The Driver combines the outputs of lidar, cameras, radar, audio, positioning systems, and maps. Cameras are strong at visual semantics; lidar provides direct three-dimensional range; radar contributes distance and motion information and can be useful in adverse weather; and audio can reveal relevant sounds outside the vehicle’s immediate visual interpretation.
This combination is called sensor fusion. It does not mean the vehicle blindly averages every sensor reading. Software must determine which observations correspond to the same object, resolve differences, account for uncertainty, and maintain a consistent model of the scene. The overlapping modalities provide redundancy: when one sensor is affected by darkness, glare, rain, fog, obstruction, or another limitation, other sensors may still contribute useful evidence.
3. Perception: turning raw data into a world model
Raw point clouds, images, radar returns, audio signals, and positioning data are not yet a driving decision. The perception system processes them into a structured representation of the environment.
That representation can include the locations and classifications of pedestrians, cyclists, cars, trucks, traffic signals, signs, curbs, lane boundaries, construction equipment, road edges, and temporary obstacles. It also needs to track how those objects are moving and how confidently the system understands each observation.
Recommended Free Tools
For example, perception may determine that an object beside the road is a cyclist rather than a stationary sign, that a light is red rather than merely reflecting sunlight, or that a temporary sign has changed the normal road arrangement. This is why autonomous driving requires more than object detection in the simple sense. The system has to interpret objects in relation to lanes, traffic rules, road geometry, and the vehicle’s intended path.
4. Prediction: estimating what other road users may do
A vehicle cannot drive safely by reacting only to where objects are at the current instant. It must also estimate what nearby road users may do next.
A pedestrian standing near a crosswalk may enter it. A cyclist may move around a parked vehicle. A car may merge, turn, stop, yield, or continue through an intersection. Waymo describes prediction as using current observations and accumulated driving experience to consider possible future paths for surrounding road users.
Prediction is uncertain by nature. The system does not need to know one guaranteed future—because there usually is no guaranteed future. It must consider plausible possibilities and select behavior that remains safe across the important ones. That is why an autonomous vehicle may slow down, pause, or leave additional space even when a dangerous event has not yet occurred. It is managing uncertainty rather than waiting for a hazard to become unavoidable.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitches5. Planning: deciding what the car should do
After localization, perception, and prediction produce a current model of the situation, the planning system determines how to proceed. It selects a route toward the destination and a specific path and speed that fit the road, traffic rules, vehicle capabilities, and surrounding activity.
Planning can be understood as several related levels:
- Route planning chooses the roads and turns that lead toward the destination.
- Behavior planning determines whether the vehicle should stop, yield, merge, change lanes, proceed through an intersection, or wait for another road user.
- Motion planning generates a precise, smooth trajectory with appropriate speed, steering, clearance, and timing.
The actual production implementation is proprietary, but this layered explanation is useful because it shows why “the car picked a route” is only a small part of driving. A route tells the vehicle which streets to use. A motion plan tells it how to navigate the next few seconds while respecting lane geometry, traffic rules, nearby objects, safety margins, and the physical limits of the vehicle.
6. Compute and vehicle control: turning a plan into motion
Onboard computing is the real-time integration point for the system. Waymo describes using powerful CPUs and GPUs, along with custom silicon in newer hardware, to process sensor inputs, identify objects, maintain the vehicle’s model of the scene, and plan movement in real time.
The computer’s output must then become physical motion. Vehicle-control systems execute commands for steering, braking, acceleration, and related functions. The system continuously observes the result of those commands and compares the vehicle’s actual position and motion with the intended trajectory.
This is a feedback loop:
- Sensors observe the vehicle and its surroundings.
- Software estimates location, objects, movement, and uncertainty.
- The planner selects a route and short-term trajectory.
- Control systems apply steering, braking, and acceleration commands.
- New observations reveal how the vehicle and environment changed.
- The plan is updated continuously.
That loop is why autonomous driving is not a prerecorded sequence of steering instructions. The vehicle must react to conditions that were not present when the map was created and correct its behavior as the real world changes.
For readers who want to study the perception–prediction–planning–control pipeline in more depth, an autonomous vehicle technology book can provide broader coverage of localization, mapping, artificial intelligence, simulation, planning, control, and safety. Such books explain the engineering concepts; they should not be treated as documentation of Waymo’s proprietary production code or exact implementation.
Why the car needs maps if it can see
Maps and sensors solve different problems:
| Detailed maps provide | Live sensors provide |
|---|---|
| Stable lane geometry and road layout | Current traffic and pedestrian positions |
| Known stop signs, traffic signals, curbs, and crosswalks | Temporary signs, construction, debris, and blocked lanes |
| A precise local reference for localization | Weather, visibility, and changing road conditions |
| Expected road structure | What other road users are doing right now |
A map can say that a lane should exist at a particular location. Sensors determine whether that lane is currently open, whether a truck is blocking it, whether a cyclist is using it, and whether roadwork has changed the normal arrangement. The Driver must reconcile the mapped expectation with the live observation rather than treating either source as complete by itself.
How Waymo tests and validates the system
Autonomous driving cannot be validated through one test type. Waymo describes combining component testing, closed-course testing, public-road driving, simulation, and system-level testing.
Component and system testing
Individual sensors, computers, software modules, and vehicle interfaces can be tested separately. Full-system testing then examines how those components work together, from sensing and perception through planning and vehicle control. Waymo’s sixth-generation materials describe validation from the component level through the complete system.
Closed-course testing
A controlled course allows engineers to stage specific situations and repeat them consistently. This is useful for evaluating behaviors such as stopping, yielding, obstacle handling, lane changes, and interactions with other road users under controlled conditions.
Public-road testing
Real roads expose the system to the variety and unpredictability of ordinary driving: unusual traffic behavior, temporary changes, local road design, imperfect visibility, and interactions that are difficult to reproduce exactly on a test track.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Simulation
Simulation allows rare, dangerous, or difficult-to-stage scenarios to be replayed and varied at scale. Engineers can test many possible combinations of road-user behavior, road layout, and system response without exposing people to the physical risk of every experiment.
Simulation is valuable, but it does not prove that a vehicle is safe in every possible circumstance. Models can omit real-world details, and unusual conditions can expose assumptions that were not represented accurately. The stronger approach is to use simulation as one part of a broader validation and safety-assurance process that also includes real-world, closed-course, component, and system-level testing.
A safety case brings this evidence together. It organizes the assumptions, hazards, mitigations, test results, operational limits, and other controls used to decide whether deployment is appropriate in a particular environment.
Can the Waymo Driver operate at night or in bad weather?
The sensor combination is designed to work beyond ideal daylight conditions. Lidar generates its own laser illumination, radar supplies distance and relative-speed information, and cameras are designed with low-light and high-dynamic-range capabilities. Overlapping sensor coverage gives the system more than one way to interpret important parts of the scene.
Waymo’s published all-weather materials describe work involving rain, fog, sandstorms, freezing temperatures, and efforts to extend capability into snowier winter environments. That should not be read as a promise of unrestricted operation in every severity of every weather event. A system may be designed and tested for a weather condition while still having operational limits based on visibility, precipitation, road traction, sensor contamination, temperature, or the specific service area.
Newer hardware descriptions also mention sensor-cleaning systems intended to help preserve performance when cameras or other sensors encounter dirt, water, snow, or road spray. Cleaning hardware reduces one problem; it does not eliminate the need for the software to recognize uncertainty and respond conservatively.
Is a human secretly driving the vehicle remotely?
Waymo describes remote assistance as contextual help, not continuous remote driving. In an unusual or difficult situation, the Driver can request additional information. A remote agent may provide advice about the context, but the autonomous system decides whether and how to use that advice.
In this model, a remote specialist is not continuously monitoring every vehicle as a substitute driver and does not routinely take over the steering wheel from thousands of miles away. Remote assistance is better understood as an additional source of information that the vehicle can consult when its own interpretation of an unusual situation needs context.
What safety evidence can—and cannot—show
Waymo publishes analyses comparing its crash rates with human-driver benchmarks. Its stated methodology uses police-reported crash information, vehicle-miles-traveled data, and comparisons aligned with the locations and conditions in which Waymo operates.
Those results should be presented with their attribution and scope. A company-reported comparison is evidence about the analyzed fleet, mileage, operating areas, time period, crash definitions, and comparison method. It is not a universal guarantee that every ride, road, weather condition, vehicle platform, or future software version will produce the same result.
Safety figures are also time-sensitive. The fleet changes, the Driver is updated, operating areas expand, mileage accumulates, and analysis methods can be refined. Readers evaluating a particular safety claim should check the date, population, geography, and conditions covered rather than treating a single percentage as a permanent property of autonomous driving.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Which cars use the Waymo Driver?
The base vehicle and the autonomous-driving system are separate parts of the overall product. The vehicle platform supplies the chassis, electric powertrain, passenger cabin, braking and steering hardware, and other automotive systems. Waymo integrates its sensors, onboard computing, software, maps, controls, and operating systems with that platform.
Waymo’s published FAQ identifies the fully electric Jaguar I-PACE and the Ojai among its fleet platforms. Separately, Waymo has described adapting its sixth-generation Driver to additional platforms, including the Hyundai IONIQ 5. The exact fleet mix and availability can change as vehicle generations and service deployments change.
For that reason, “the Google self-driving car” is an imprecise description. There is no single consumer car that contains the entire system. The more accurate name for the technology is the Waymo Driver, integrated with a particular vehicle platform and deployed with maps, software, validation, and operational support for a defined environment.
What the Waymo Open Dataset does—and does not—show
Waymo makes an Open Dataset available for research. It includes several kinds of material, including:
- Perception data with sensor information and labels.
- Motion data with object trajectories and three-dimensional maps.
- End-to-end driving data with camera imagery and routing instructions.
The dataset helps researchers study the types of perception, tracking, mapping, and motion problems involved in autonomous driving. It is not the complete production Driver. Waymo describes the public dataset as only a fraction of the data and capabilities used by its operational system, so it should not be mistaken for an open-source copy of Waymo’s software or hardware.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallCommon misconceptions about Google’s self-driving car
“It is just a camera.”
No. The system uses multiple modalities, including lidar, cameras, radar, and, in newer hardware descriptions, external audio receivers. The purpose is to combine complementary information and provide redundancy.
“GPS tells it how to drive.”
GPS is not the complete localization or navigation system. The Driver matches detailed maps with live sensor observations to establish a precise local understanding of the road and its current conditions.
“A remote human drives whenever the car gets confused.”
Waymo describes remote assistance as advice or contextual information requested by the Driver, not continuous teleoperation. The autonomous system remains responsible for deciding whether and how to use that information.
“It learns only by driving on public roads.”
Development and validation also involve simulation, closed-course testing, component testing, system-level testing, and public-road operation. Each method supplies evidence that the others cannot provide alone.
Recommended Free Tools
“Any lidar, radar, or dashcam can turn a normal car into a Waymo.”
No. The Waymo Driver is an integrated hardware-and-software platform. It depends on sensor integration, onboard computing, maps, machine-learning and planning software, vehicle controls, validation, safety processes, and operational support. A consumer accessory may use a related technology without reproducing the complete system.
“Self-driving and driver assistance are the same thing.”
They are not. Driver-assistance systems require a human driver to remain responsible and attentive. Waymo’s autonomous ride-hailing operation is designed to perform the driving task within its approved operating conditions, without treating a human passenger as the backup driver. That does not mean it can drive anywhere, in every weather condition, or under every circumstance.
The simplest way to understand how it works
Google’s original self-driving project—and the Waymo Driver that followed—works as a coordinated stack:
- Maps describe the expected road structure.
- Localization matches live observations to that structure.
- Lidar, cameras, radar, and audio gather overlapping information about the current scene.
- Perception turns that information into a model of objects, lanes, signs, signals, and road conditions.
- Prediction estimates how nearby road users may move.
- Planning chooses a route, behavior, speed, lane, and trajectory.
- Control applies steering, braking, and acceleration commands.
- Feedback and validation continually evaluate the result and improve confidence in the system.
The impressive part is not a single magical sensor or artificial-intelligence feature. It is the integration of many systems that must work together in real time, with enough redundancy, testing, and operational discipline to handle the uncertainty of public roads.
Free tools Windows power users keep installed
One-click scans. No signup required.
Frequently Asked Questions
What happened to Google’s self-driving car project?
Google’s original self-driving project became Waymo, an Alphabet company focused on autonomous-driving technology and autonomous ride-hailing. The current technology is called the Waymo Driver.
Is a human remotely driving the Waymo vehicle?
No. Waymo describes remote assistance as contextual advice requested by the autonomous system in unusual situations, not continuous remote driving or routine teleoperation.
Why does a self-driving car need detailed maps?
Maps provide a detailed reference for stable features such as lanes, signals, curbs, and crosswalks. Sensors detect changing conditions such as traffic, pedestrians, construction, debris, and temporary signs.
What sensors does the Waymo Driver use?
Waymo describes a sensor suite that includes lidar, cameras, radar, and newer external audio receivers. Each modality contributes different information and helps compensate for limitations in the others.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Can the Waymo Driver operate in darkness and bad weather?
Waymo’s published materials describe work in rain, fog, sandstorms, freezing temperatures, and snowier conditions, but that does not mean unrestricted operation in every severity of every weather event. Service conditions and system limits still matter.
The Bottom Line
Bottom line: Google’s self-driving car is now best understood as the Waymo Driver, an integrated autonomous-driving system rather than a single car model. It uses detailed maps and multiple live sensors to locate the vehicle and understand its surroundings, predicts what nearby road users may do, plans a safe trajectory, and continuously controls the vehicle while being evaluated through simulation, closed-course testing, and public-road validation.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
