The promise of autonomous freight has been circling the logistics industry for nearly a decade, but 2026 marks a genuine inflection point. Several self driving truck manufacturers have moved past pilot programs into revenue-generating, driverless commercial operations on specific highway corridors. Others have retrenched, pivoted, or shut down entirely. For fleet operators, logistics planners, and EV-curious readers tracking the electrification and automation of heavy transport, the landscape demands a clear-eyed breakdown — not hype.
This guide profiles the major self driving truck manufacturers operating or deploying in the U.S. right now, their technology stacks, where their trucks actually run without a human behind the wheel, what the safety data shows, and the regulatory and physical limitations that still constrain the industry.
Understanding Autonomy Levels in Trucking: SAE Level 4 vs. Level 2
Before comparing manufacturers, it is essential to distinguish between driver-assist and true driverless operation. The SAE International J3016 standard defines six levels of driving automation, and the trucking industry currently operates across three of them:
| SAE Level | Name | Who Drives? | Trucking Example |
|---|---|---|---|
| Level 2 | Driver Assistance | Human driver, hands on wheel | Highway cruise + lane centering (e.g., Detroit Assurance 5.0 in Freightliner Cascadia) |
| Level 3 | Conditional Automation | System drives in limited conditions; human must take over on request | Torc Robotics / Daimler Truck testing (not in commercial driverless service) |
| Level 4 | High Automation | System drives without human intervention within its Operational Design Domain (ODD) | Aurora Driver, Kodiak Driver, Waymo Via — on geofenced highway corridors |
| Level 5 | Full Automation | System drives everywhere, all conditions | Does not exist in commercial trucking |
The critical takeaway: every commercial autonomous truck in revenue service today is SAE Level 4 — meaning it operates without a human driver, but only within a defined geofence, on mapped highway routes, and under specific weather and time-of-day conditions. No self driving truck manufacturer offers Level 5 capability, and none claim to. When companies reference "self-driving," they mean Level 4 within an Operational Design Domain.
The Major Self Driving Truck Manufacturers
The field has consolidated significantly since 2023. Here are the companies currently deploying or operating autonomous Class 8 trucks on U.S. roads.
Aurora Innovation
Aurora is arguably the most prominent self driving truck manufacturer in commercial operation. The company launched its driverless freight service — branded Aurora Connect — on a Dallas-to-Houston corridor in late 2024, and by early 2026 has expanded to include routes between Dallas and Fort Worth, and a Dallas-to-El Paso lane operated in partnership with FedEx and Uber Freight.
Aurora's trucks are built on Volvo VNL 860 and PACCAR Peterbilt 579 platforms, retrofitted with the Aurora Driver hardware suite. The company has logged over 4 million autonomous miles on public roads as of Q1 2026, with a growing portion of those miles fully driverless (no safety operator in the cab). Aurora's business model is "Driver-as-a-Service" — fleets pay per mile rather than purchasing the hardware outright.
Kodiak Robotics
Kodiak focuses on long-haul highway autonomy and has carved out a niche in the Texas freight triangle (Dallas–Houston–San Antonio). The company operates a fleet of autonomous Peterbilt 579 trucks equipped with its fourth-generation sensor suite. As of 2026, Kodiak reports approximately 2.5 million autonomous miles, with driverless commercial freight deliveries running daily on I-35 and I-45.
Kodiak differentiates on its modular "sensor pod" design — a roof-mounted unit containing lidar, cameras, and radar that can be swapped in under two hours, simplifying maintenance. The company has partnerships with Atlas Energy Solutions (hauling oilfield equipment) and has begun limited operations in the Southeast.
Waymo Via
Waymo's autonomous trucking division, Waymo Via, leverages the same core perception and planning stack as its robotaxi fleet. Waymo Via trucks operate primarily in Texas and have tested in Arizona. The company's approach integrates deeply with its fifth-generation Waymo Driver hardware — a dense array of lidar, cameras, and radar units.
Waymo Via has been more selective about scaling than Aurora, focusing on proving reliability metrics before expanding route coverage. The company has partnered with J.B. Hunt and UPS for freight hauling, with driverless operations on defined highway segments. Waymo Via's total autonomous trucking miles are not publicly separated from its robotaxi mileage, but the combined Waymo fleet has surpassed 40 million fully autonomous miles across all vehicle types.
TuSimple (Reoriented)
TuSimple was once a leading self driving truck manufacturer, but the company wound down its U.S. autonomous trucking operations in late 2023 following management turmoil and SEC scrutiny. As of 2026, TuSimple has pivoted to the Asia-Pacific market, with autonomous truck pilots in China and Australia. It is no longer a factor in U.S. autonomous freight, though its historical data and patents remain relevant to the industry.
Other Notable Players
- Torc Robotics (a Daimler Truck subsidiary) — developing Level 4 hub-to-hub autonomy for Freightliner platforms; testing on public highways in New Mexico and Virginia, but not yet in driverless commercial service as of early 2026.
- Pronto.ai — offers a Level 2 driver-assist system (not Level 4) for commercial trucks, focused on highway safety features like automatic emergency braking and lane-keeping.
- Gatik — operates autonomous box trucks (Class 5-6, not Class 8) for middle-mile logistics between Walmart distribution centers and stores, primarily in Texas, Arkansas, and Louisiana. These run driverless on fixed, short routes.
Where Autonomous Trucks Actually Operate Today
Driverless Class 8 trucks are not running nationwide. Every self driving truck manufacturer limits its operations to carefully mapped, geofenced highway corridors. Here is the current service footprint:
| Manufacturer | Active Corridors | Primary Highways | Driverless Status |
|---|---|---|---|
| Aurora | Dallas–Houston, Dallas–Fort Worth, Dallas–El Paso | I-45, I-35, I-20, I-10 | Commercial driverless (since late 2024) |
| Kodiak | Dallas–Houston, Dallas–San Antonio, expanding Southeast | I-35, I-45, I-10 | Commercial driverless (since 2024) |
| Waymo Via | Texas (Dallas/Houston area), Arizona testing | I-35, I-45 | Commercial driverless (limited volume) |
| Gatik | Dallas–Fort Worth, Bentonville AR, New Orleans | Surface streets + short highway segments | Commercial driverless (middle-mile box trucks) |
| Torc Robotics | Albuquerque NM, Virginia (testing only) | I-25, I-81 | Safety driver present — not yet driverless |
Texas dominates autonomous trucking for a reason: wide multi-lane interstates, favorable regulatory environment, high freight volume, and relatively mild winter weather. The state's HB 2833, signed in 2023, established a clear legal framework for autonomous commercial vehicles, removing ambiguity about liability and permitting.
Technology Stack: How Self Driving Trucks See the Road
Every Level 4 autonomous truck relies on a layered sensor suite, high-definition maps, and redundant compute. Here is the general architecture shared across major manufacturers, with company-specific variations noted:
Perception Sensors
- Lidar (Light Detection and Ranging): Primary long-range sensor. Aurora uses a proprietary lidar called "FirstLight" (FMCW technology, 400+ meter range). Kodiak uses a combination of Luminar and Hesai lidars with a 250+ meter range. Lidar provides precise 3D point clouds regardless of lighting conditions.
- Cameras: Multiple high-resolution cameras (typically 8-14 per truck) covering 360° around the vehicle. Used for lane-line detection, traffic sign recognition, and object classification. Waymo Via uses custom-designed cameras integrated into its fifth-generation hardware.
- Radar: Forward and side-facing radar units (77 GHz) for velocity detection of other vehicles and redundancy in heavy rain or fog where lidar performance degrades.
Compute and Planning
- Aurora: Custom "Aurora Orion" compute platform with redundant NVIDIA-based processing. Runs a unified perception-planning model trained on billions of simulated and real-world miles.
- Kodiak: NVIDIA DRIVE Orin-based compute with a proprietary "Virtual Safety Driver" planning stack. Emphasizes a modular, replaceable hardware approach.
- Waymo Via: Custom sixth-generation Waymo Driver compute, shared with the robotaxi fleet. Benefits from the largest real-world autonomous driving dataset in the industry.
HD Mapping and Geofencing
All manufacturers rely on centimeter-accurate high-definition maps of their operating corridors. These maps include lane geometry, merge points, construction-zone profiles, and speed-limit data. Trucks will not operate outside these pre-mapped geofences — if a route deviation is required (e.g., a detour due to an accident), the system will pull over and request remote human guidance or wait for the corridor to reopen.
Safety Data: What the Numbers Actually Show
Autonomous truck safety is the most scrutinized aspect of the industry. Here is what the available data indicates as of early 2026:
- Aurora reported a disengagement rate (human intervention or system-initiated pull-over) of approximately 1 intervention per 12,000 autonomous miles during its first year of driverless operations (late 2024–late 2025). The company has not reported any at-fault collisions during driverless service.
- Kodiak reports a disengagement rate of roughly 1 per 10,000 miles, with the majority of disengagements triggered by construction-zone reconfiguration or sensor occlusion from road debris — not by planning failures.
- Waymo Via does not publish truck-specific disengagement data separately, but Waymo's aggregate robotaxi data shows a crash rate approximately 85% lower than human benchmarks for comparable urban driving, based on 40+ million autonomous miles. Highway trucking data is more limited.
- No fatal at-fault collisions involving a Level 4 autonomous Class 8 truck in commercial service have been reported by any manufacturer as of Q1 2026.
For context, the National Highway Traffic Safety Administration (NHTSA) reports that large trucks were involved in approximately 5,800 fatal crashes in 2023, with human error cited as a factor in the vast majority. The autonomous trucking industry argues that even early Level 4 deployments, limited to well-mapped highway corridors, can eliminate fatigue-related and distraction-related incidents — which account for an estimated 30-40% of trucking fatalities.
However, the total autonomous trucking mileage remains tiny compared to the 300+ billion miles driven annually by human truckers in the U.S. Statistical significance for safety comparisons will require tens of millions more autonomous miles.
Cost Structure: Autonomous Trucking vs. Human-Driven Freight
Autonomous trucking is not yet a consumer-facing service like a robotaxi — you cannot hail an Aurora truck on an app. The pricing model is business-to-business, structured as per-mile freight rates charged to shippers and carriers. Here is how the economics compare:
| Cost Component | Human-Driven Truck | Autonomous Truck (Level 4) |
|---|---|---|
| Driver wages + benefits | $0.60–$0.80/mi | $0.00 (no in-cab driver) |
| Autonomous system fee (hardware + software + remote ops) | N/A | $0.40–$0.70/mi (varies by provider) |
| Fuel / energy | $0.50–$0.70/mi (diesel) | $0.50–$0.70/mi (diesel; EV trucks emerging) |
| Maintenance | $0.15–$0.20/mi | $0.18–$0.25/mi (additional sensor maintenance) |
| Insurance | $0.05–$0.08/mi | $0.04–$0.06/mi (lower liability, but evolving) |
| Estimated total per mile | $1.30–$1.78 | $1.12–$1.71 |
The cost advantage of autonomous trucking is real but modest at current scale. The savings come from eliminating driver wages (the single largest variable cost in trucking) and enabling near-24/7 operation — autonomous trucks do not require rest breaks under current federal Hours-of-Service exemptions for driverless vehicles. However, the autonomous system fee partially offsets those savings, and sensor maintenance adds cost. As manufacturing scales and system costs decline, the per-mile gap is expected to widen in autonomy's favor by 2028-2030.
Regulatory Landscape: Who Approves Autonomous Trucks?
Autonomous trucking regulation in the U.S. is a patchwork of state and federal authority:
- Federal (NHTSA / FMCSA): The National Highway Traffic Safety Administration sets vehicle safety standards (FMVSS). The Federal Motor Carrier Safety Administration regulates commercial driver qualifications, hours of service, and carrier safety ratings. FMCSA has issued exemptions allowing driverless commercial operations under specific conditions but has not created a comprehensive federal framework for autonomous trucks.
- State level: States control whether autonomous trucks can operate on their roads. As of 2026, Texas, Arizona, Florida, Georgia, and Arkansas have permissive frameworks for Level 4 autonomous commercial vehicles. California banned autonomous truck testing on public roads in 2023 (via DMV action against TuSimple), and the state remains restrictive. New York, New Jersey, and most Northeastern states do not permit driverless Class 8 operations.
- Insurance and liability: In driverless operations, the autonomous system provider (Aurora, Kodiak, Waymo Via) typically assumes liability for at-fault incidents — a significant shift from traditional carrier liability models.
The lack of federal preemption means self driving truck manufacturers must seek approval state by state, which is a major barrier to interstate route expansion. A truck operating driverless from Dallas to Houston cannot continue driverless into Louisiana or New Mexico without separate state authorization.
Current Limitations: What Autonomous Trucks Cannot Do
Despite real commercial progress, Level 4 autonomous trucks face significant constraints that limit where, when, and how they operate:
- Highway-only operation: Current systems handle divided, multi-lane interstates. They cannot navigate urban surface streets, loading docks, truck stops, or warehouse yards. Freight is typically transferred at "transfer hubs" near highway exits, where a human driver handles the first and last mile.
- Weather restrictions: Heavy rain, snow, ice, and dense fog degrade lidar and camera performance. Most manufacturers suspend driverless operations in precipitation exceeding moderate rain. Snow and ice are entirely off-limits — which is why Texas and Arizona dominate deployment.
- Construction zones: Unmapped lane shifts, temporary barriers, and flagging operations remain challenging. While systems can handle minor, pre-mapped construction, unexpected reconfigurations trigger a pull-over and remote-assist request.
- Geofence rigidity: Trucks cannot deviate from pre-mapped corridors. An accident closure requiring a detour outside the geofence means the truck stops until the route reopens or a remote operator provides an approved alternate path.
- No night operations (some providers): While lidar works in darkness, some manufacturers restrict nighttime driverless operations during early deployment phases to reduce edge-case risk.
- Hazardous materials: Federal regulations currently prohibit autonomous vehicles from transporting placarded hazardous materials. This rules out a significant segment of freight.
- Weight and configuration limits: Most systems are validated for standard 53-foot dry van and refrigerated trailers. Oversized loads, doubles/triples, and flatbed configurations with shifting cargo are not supported.
The Electrification Intersection: Autonomous + Electric Trucks
A natural question is whether autonomous trucks are also electric. As of 2026, the answer is mostly no — the vast majority of Level 4 autonomous trucks run on diesel. The reason is practical: long-haul routes (300-600 miles) exceed the current range of Class 8 battery-electric trucks, and the charging infrastructure for heavy trucks is still in early development.
However, the intersection is coming. Aurora has announced a partnership to integrate its Driver system into the Volvo VNR Electric for short-haul and regional routes. Kodiak has tested its system in a PACCAR battery-electric Peterbilt prototype. Gatik's box trucks, which run shorter routes, are more compatible with electrification and the company has tested electric platforms.
The convergence of autonomy and electrification is most likely to appear first in middle-mile logistics (50-150 mile routes with depot charging) rather than long-haul interstate freight.
Frequently Asked Questions
Is an autonomous truck truly self-driving, or is it driver-assist?
The trucks from Aurora, Kodiak, and Waymo Via operating in commercial service are SAE Level 4 — meaning they drive themselves without a human in the cab, but only on pre-mapped highway corridors under defined conditions. This is fundamentally different from Level 2 driver-assist systems like Autopilot or BlueCruise, which require a human driver to remain engaged at all times. No Level 5 (drive-anywhere) autonomous truck exists.
Where can autonomous trucking services be used right now?
Driverless Class 8 freight is currently available on specific Texas highway corridors: Dallas–Houston (I-45), Dallas–San Antonio (I-35), and Dallas–El Paso (I-20/I-10). Gatik operates autonomous box trucks on shorter middle-mile routes in Texas, Arkansas, and Louisiana. Expansion into additional states is underway but requires state-by-state regulatory approval.
How safe are self-driving trucks compared to human drivers?
Early data is promising but limited by sample size. Major manufacturers report disengagement rates of approximately 1 per 10,000-12,000 miles and no at-fault fatal collisions in driverless service. For context, human-driven large trucks are involved in roughly 5,800 fatal crashes per year nationally. However, autonomous trucks have collectively driven only a few million driverless miles versus 300+ billion human-driven truck miles annually — statistical parity in safety data will take years to establish.
What happens when an autonomous truck encounters a problem it cannot handle?
The system executes a "minimal risk condition" — typically pulling onto the shoulder or into a safe lane position and stopping. A remote operations center is alerted, and a human remote-assistance operator can evaluate the situation and provide guidance (e.g., approving a lane change around an obstacle). The truck does not make decisions outside its validated capability set.
Will autonomous trucks replace human truckers?
Not in the near term. The current Level 4 systems handle only the highway portion of freight movement — the "middle mile" between transfer hubs. Human drivers still handle first-mile pickup, last-mile delivery, urban navigation, dock operations, and routes outside geofenced corridors. Industry analysts project that autonomous trucking will initially address the long-haul driver shortage (estimated at 80,000+ drivers by ATA figures) rather than displace existing workers. The timeline for broader displacement depends on technology maturation, regulatory expansion, and infrastructure development over the next decade.
Which self driving truck manufacturer should fleet operators evaluate first?
For fleets operating freight on Texas I-35 and I-45 corridors, Aurora and Kodiak both offer commercial driverless service today and should be evaluated based on per-mile pricing, freight type compatibility, and integration with existing fleet management software. For fleets outside Texas, the practical answer is to monitor expansion announcements and engage in pilot discussions with providers planning Southeast or Midwest corridor launches in 2026-2027.
