NVIDIA announced DRIVE Constellation as “now available” on March 18, 2019. That was a launch-era availability claim, not confirmation that the platform can still be obtained in 2026. NVIDIA’s current DRIVE downloads page does not establish Constellation-specific access or support. The system itself was a two-server simulation platform designed to test autonomous-driving software with simulated sensor data and a physical DRIVE AGX Pegasus computer in the loop.
What was NVIDIA DRIVE Constellation?
DRIVE Constellation was a data-center platform for testing autonomous-driving systems in simulated driving conditions. NVIDIA announced it on March 27, 2018, and said it was available on March 18, 2019, at its GPU Technology Conference. In the 2019 announcement, NVIDIA described the cloud-based platform as enabling “millions of miles” of virtual driving; in 2018, the company had claimed virtual testing at the scale of “billions of miles.” Those are NVIDIA’s launch-era descriptions, not independently verified customer mileage totals. NVIDIA’s 2019 announcement and 2018 introduction provide the historical context.
How did the two-server simulation work?
The 2019 description paired two servers to form a feedback loop: one generated the simulated world and sensor data, while the other ran the vehicle computer and software being tested.
| Part | Role in the loop |
|---|---|
| Simulator server | Used NVIDIA GPUs and DRIVE Sim to model a virtual car in a virtual world and generate simulated sensor output. |
| Vehicle server | Contained a DRIVE AGX Pegasus computer, which processed the simulated inputs and ran the vehicle software. |
| Feedback | The vehicle computer’s decisions went back to the Simulator, which updated the virtual driving situation for the next cycle. |
NVIDIA characterized this as hardware-in-the-loop testing with bit-accurate and timing-accurate behavior. In its March 2018 announcement, NVIDIA said the loop ran 30 times per second. That frequency is a launch-era specification, not a claim about current configurations. NVIDIA’s 2018 announcement
Recommended Free Tools
#1 Best Overall
- Includes OBD2 Cable & Fuse – Comes with a ready-to-use OBD2 cord and a built-in automotive fuse for safe, reliable vehicle connection.
- 3.3V or 5V Logic Compatible – Works seamlessly with ESP32, Arduino, Raspberry Pi, STM32, Teensy, and more.
- Automotive-Grade Protection – Built-in power regulation, reverse-polarity protection, and noise filtering ensure stable, safe readings from any 12V vehicle.
- Supports Major OBD-II Protocols – Works with ISO9141, ISO14230 (KWP2000) for K-Line vehicles and ISO15765-4 CAN for modern CAN Bus systems (11-bit & 29-bit IDs).
What hardware-in-the-loop meant here
In hardware-in-the-loop testing, the software under test runs on actual target hardware rather than only inside a software simulation. For Constellation, the DRIVE AGX Pegasus computer processed the simulator’s sensor inputs and returned its driving decisions to the simulated environment. That let developers exercise the vehicle computer and software in a controlled, repeatable virtual loop without treating a software-only test as equivalent to running the target hardware.
NVIDIA’s 2018 launch materials described simulated cameras, lidar and radar, along with adjustable weather, lighting, road surfaces and terrain. They also described scripting routine as well as rare or dangerous scenarios, including conditions such as glare and limited visibility at night. These are vendor descriptions of intended simulation capabilities; they are not independent evidence that a system passed safety validation or would perform safely on public roads. NVIDIA’s 2018 announcement
Rank #2
- The CAN-BUS Shield compatible with arduino or Redboard can be provided with CAN-BUS capabilities and allows you to hack your vehicle.
- This shield allows you to poll the ECU for information including coolant temperature, throttle position, vehicle speed, and engine rpms. You can also store this data or output it to a screen to make an in-dash project.
- The CAN-BUS Shield Features: CAN v2.0B up to 1 Mb/s. High speed SPI Interface (10 MHz) Standard and extended data and remote frames. CAN connection via standard 9-way sub-D connector. Power can supply to Arduino by sub-D via resettable fuse and reverse polarity protection.
- It uses the Microchip MCP2515 CAN controller with the MCP2551 CAN transceiver. CAN connection is via a standard 9-way sub-D for use with OBD-II cable. Ideal for automotive CAN application. The shield also has a uSD card holder, serial LCD connector and connector for an EM506 GPS module.
- Note: A DB9 Cable is not included with this shield.----Note: This product is a collaboration with SK Pang Electronics. A portion of each sales goes back to them for product support and continued development.
What the historical architecture says about scale
NVIDIA’s 2018 Autonomous Driving Reference Architecture describes a Constellation POD design for data-center deployment. Its examples include a rack holding four Constellation systems and eight storage nodes, eight camera channels over GMSL2, and 1 GbE connections for radar and lidar systems. These details describe that historical architecture, not current installation requirements.
The same document gives example sizing calculations rather than measured outcomes or current recommendations: 2,000 hours of raw data and 20 Constellation systems for a stated 100-hour turnaround target; and 20,000 hours of raw data and 200 systems for a ten-car development program under the document’s stated turnaround assumptions. The figures illustrate how data volume and desired turnaround could drive infrastructure needs in those examples. They should not be used as a present-day sizing guide.
Free tools Windows power users keep installed
One-click scans. No signup required.
Simulation supports safety work; it does not replace road testing
Simulation can make it practical to repeat scenarios and explore combinations that are uncommon, difficult or unsafe to reproduce on a public road. But simulation alone does not establish that an automated-driving system is safe for deployment. NVIDIA’s Self-Driving Safety Report describes combining actual road miles with simulated miles. That is the more useful framing: simulation is one part of a validation program, alongside real-world testing and other safety work.
Who was associated with Constellation at launch?
NVIDIA’s March 2019 release named Toyota Research Institute-Advanced Development (TRI-AD) as the first customer. It said TÜV SÜD was using the platform to formulate self-driving validation standards. NVIDIA also described Cognata’s traffic and scenario simulation as supportable on Constellation and named IPG Automotive as an ecosystem partner whose CarMaker software could create virtual vehicle prototypes and model subsystem responses. These are historical claims from the launch announcement; they do not establish present customer use, compatibility or commercial relationships. NVIDIA’s 2019 announcement
Rank #4
- Engine Management Systems
Is DRIVE Constellation still available?
Current availability, support and commercial terms could not be verified from NVIDIA’s published developer information. Its DRIVE Downloads page discusses broader autonomous-vehicle data, training and simulation workflows, as well as developer-program membership and licensing, but does not establish whether DRIVE Constellation itself can currently be accessed or supported. The 2019 “now available” statement should therefore be read as historical, not as a present-day offer.
What to evaluate when comparing simulation approaches
The available Constellation materials suggest useful questions for a professional evaluation, but they do not establish a current shortlist or rank competing products. Ask:
Quick Recap
Best Value
- Support CAN V2.0B technical specification, communication rate 1Mb/S.
- 0~8 bytes long data field, standard frame, extended frame and remote frame.
- Module 5V DC power supply, SPI interface protocol control, 120 ohm terminating resistor, impedance matching, guaranteed drive capability, long-distance data transmission to prevent signal emissions.
- Module size: 44mm x 28mm, centering distance of the positioning screw hole: 23mm x 38mm.
- Operating current: typical value 5mA, standby current 1 microamperes, except for the power indicator. Working temperature: industrial grade -40 ° C to 85 ° C.
- Does testing run software-only, or on the target vehicle computer in the loop?
- Which sensor types are simulated, and what level of sensor realism is needed?
- Can teams create, repeat and vary the scenarios relevant to their validation goals?
- How are traffic models and virtual vehicle or subsystem models integrated?
- What throughput and turnaround time does the program require?
- What data-center hardware, networking, power and cooling would the architecture require?
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.
