The dispatch supervisor at a 90-truck dry-van carrier in Fort Worth pulled up her TMS (transportation management system) on a Monday morning in March 2026 and saw something she had never seen before: two loads on her board tagged with a lane type she did not recognize. Her team had just booked autonomous capacity through the McLeod TMS integration for a pair of Dallas-to-El Paso corridor runs. The trucks were Aurora vehicles operating under SAE (Society of Automotive Engineers) Level 4 autonomy, meaning they required no human driver intervention inside their approved operational design domain (ODD), the geographic and environmental envelope within which the system is certified to operate without human backup. The loads were booked, confirmed, and ready to dispatch. The problem was that nobody in that office had a protocol for what happened next. Who monitored the truck? What happened at the El Paso relay point? What if the autonomous vehicle (AV) hit a weather condition outside its ODD? The loads were real. The plan was missing.

The Mixed-Fleet Reality in 2026

The autonomous long-haul market reached $2.7 billion in 2024 and is growing at approximately 32 percent CAGR (compound annual growth rate) toward a projected $42.6 billion by 2034. That is not a forecast about the distant future. Aurora has logged more than 250,000 driverless commercial miles and made its capacity bookable through the McLeod TMS integration that now serves more than 1,200 fleets. The FMCSA (Federal Motor Carrier Safety Administration) is actively updating its hours-of-service (HOS) rules to address driverless operations, recognizing that a vehicle without a human driver does not accrue HOS time the way a human driver does. The regulatory infrastructure, the technology, and the TMS wiring are all converging at the same moment.

What that means for the fleet strategist is that autonomous capacity is no longer a pilot experiment you opt into for a press release. It is a category of freight capacity that is available for booking today, that competes on certain lanes with human-driven capacity, and that requires an operating model your dispatch office, your safety function, and your customer-service team are probably not yet ready to execute. The word "mixed fleet" describes the period every carrier will live in for the foreseeable future: some loads moving on autonomous vehicles, others on human-driven trucks, on the same network, under the same TMS, requiring coordinated management that neither the AV vendors nor the TMS platforms have fully solved for you.

The strategic question is not whether your fleet will encounter autonomous capacity. It already has, or it will within the next operating cycle. The question is whether you have thought through the operating plan before you book the first load, or whether you will discover the gaps, like that Fort Worth dispatcher, when the truck is already rolling and the protocols do not exist yet.

Understanding the ODD and What It Means for Lane Selection

The operational design domain is the single most important concept for a fleet manager planning a mixed operation. An autonomous vehicle certified for SAE Level 4 operation is not a truck that can go anywhere. It is a truck that can operate safely without human intervention within a defined envelope of conditions: specific geographic corridors, weather ranges (typically no heavy precipitation, defined visibility minimums), road types (primarily interstates with consistent lane markings and grade standards), daylight or lighting conditions, and speed ranges. Outside that envelope, the system is not certified for driverless operation, and the carrier is responsible for ensuring the load is never dispatched into conditions the AV cannot handle.

Aurora's current primary ODD covers specific interstate corridors in Texas, the broader Sun Belt, and selected Midwest routes. That coverage is expanding, but in 2026 it is not a national network. This matters for lane selection because a shipper asking for a Dallas-to-Chicago load cannot be served with Aurora autonomous capacity if the northern segment of that lane falls outside Aurora's certified corridor. A carrier that books the load as autonomous and then discovers mid-route that the ODD does not cover the delivery zone has a serious problem: the truck either needs a driver inserted at a transfer point, or the load is late, or both.

The practical discipline is lane mapping. Before a carrier commits to adding autonomous capacity to its mix, the fleet manager needs to map the carrier's existing lanes against the AV vendor's published ODD, identify the lanes where autonomous coverage is complete (origin to destination within the ODD, with no ODD exit required), and identify the lanes that require a human-driven relay segment. That mapping exercise produces three categories of lanes.

The first category is full-autonomous lanes: the complete origin-to-destination movement is inside the ODD, no relay is needed, and the carrier can dispatch an AV and manage it remotely through the TMS. These are the lowest-complexity lanes to start with, and they should be the first lanes a carrier targets for autonomous deployment.

The second category is hub-and-spoke lanes: the autonomous segment covers the long middle leg (typically the 400-to-700-mile interstate corridor), with human-driven first-mile and last-mile segments at each end. The AV is handed off at a relay or transfer hub, a human driver completes the urban pickup or delivery segment, and the AV repositions or takes the next available load. These lanes require transfer hub coordination, handoff protocols, and a clear definition of which entity (the carrier, the AV vendor, or a third party) manages the hub operation.

The third category is mixed-constraint lanes: the lane has characteristics (urban last mile, weather exposure, specialized equipment, customer-specific terminal requirements) that make autonomous coverage either impossible or commercially impractical in the current technology cycle. These lanes stay human-driven, full stop, and the carrier does not waste time evaluating autonomous capacity for them in this planning cycle.

The lane mapping discipline is not a one-time exercise. As Aurora and other AV operators expand their ODDs, the map changes. A lane that was category three in early 2026 may become category two by the end of the year. The fleet manager who has done the mapping and can quickly re-evaluate when an ODD announcement comes out is in a structurally better position than one who is starting from scratch every time a vendor sales rep calls.

The Operating Plan: Six Components Every Mixed Fleet Needs

The Fort Worth dispatcher's problem was not a booking problem. The booking worked fine. It was an operating plan problem. The carrier had no written protocol for any of the six components a mixed fleet requires. Producing those six components before the first autonomous load rolls is the fleet strategist's core deliverable in this transition period.

Component One: Dispatch Routing Rules

The TMS needs to know how to offer autonomous capacity to dispatchers, how to surface ODD compliance information at the point of load assignment, and what override logic applies when an autonomous slot is available but the dispatcher has a reason to use human capacity instead. This is a TMS configuration question as much as a policy question, and it requires coordination between the fleet's TMS administrator and the AV vendor's integration team. The McLeod-Aurora integration passes available AV capacity into the TMS's load board view, but the rules for when to offer that capacity, and at what price relative to human-driven options, need to be set by the carrier.

A practical starting point is to configure the TMS to flag lanes as "autonomous eligible" based on the ODD mapping described above, and to require a dispatcher confirmation step before any autonomous assignment is committed. That confirmation step is not bureaucracy; it is the human-in-the-loop gate that catches the load that was mis-categorized, the weather event that changed the ODD compliance picture overnight, or the customer who added a delivery requirement that takes the lane out of autonomous eligibility. The confirmation step also creates an audit trail showing that a human reviewed and approved the autonomous assignment, which matters for insurance, customer contracts, and any future regulatory inquiry.

Component Two: Safety and Exception Protocols

An autonomous truck on a long-haul interstate is not being supervised by the driver who usually answers the carrier's phone. The remote monitoring function replaces the driver as the first point of contact for exceptions, and the carrier needs to decide, in writing, how that monitoring works. In Aurora's model, Aurora's own operations center provides the remote monitoring layer and handles the technology-level exceptions (the system detecting a sensor anomaly, a weather reading outside ODD parameters, a route deviation). The carrier's responsibility is to maintain a point of contact for the business-level exceptions: a shipper calling about a delivery window, a consignee with a gate that requires advance notification, a customer service situation that the AV vendor's operations center has no authority to resolve.

The carrier also needs a written protocol for the escalation scenarios. What happens if the AV system determines it cannot complete the route under safe conditions and pulls over to a designated safe harbor? What is the carrier's responsibility for arranging a tow or a transfer? Who notifies the shipper? What is the service failure documentation process for a load that arrives late because of an AV system suspension? These scenarios are rare, but they are not hypothetical, and a carrier without written protocols for them will improvise badly under pressure.

Component Three: Transfer Hub and Handoff Logistics

Hub-and-spoke lanes require physical infrastructure: a location where the AV can drop its trailer, a human driver can pick up the trailer for the local delivery, and the inbound AV can be repositioned or loaded with the next available outbound. For carriers operating in lanes where Aurora's transfer hub network aligns with their freight patterns, this is primarily a scheduling and communication exercise. For carriers whose lanes require a transfer point that Aurora does not yet serve with a designated hub, the carrier is responsible for identifying or contracting the intermediate facility.

The handoff itself requires a defined DVIR (driver vehicle inspection report) process. When a human driver picks up a trailer that was hauled by an autonomous vehicle, who is responsible for the DVIR? Under current FMCSA regulations, the human driver who takes possession of the trailer before operating it on the road is responsible for the DVIR inspection. The carrier's protocol needs to specify that the human driver at the transfer hub completes a full DVIR before departing, that the inspection is logged in the ELD (electronic logging device) record, and that any pre-existing damage noted at the hub is documented so the carrier is not liable for damage that occurred during the autonomous segment.

Component Four: Insurance and Liability Structure

Autonomous freight introduces a liability structure that differs from human-driven operations in ways that are still being worked out by insurers, carriers, and AV vendors. In a human-driven accident, the liability analysis centers on the driver's actions and the carrier's responsibility for the driver. In an autonomous incident, the liability may involve the AV vendor (for a technology failure inside the system's certified ODD), the carrier (for dispatching the load into conditions outside the ODD, or for failing to maintain the trailer equipment that the AV was hauling), or a third party. The carrier's existing commercial auto policy was almost certainly written without autonomous operations in mind, and the carrier needs to confirm, in writing, with its insurance broker, how autonomous loads are covered before the first load moves.

The AV vendor carries its own liability coverage for technology failures within the ODD, and that coverage is part of the commercial arrangement between the carrier and the vendor. But the carrier still has exposure for equipment it owns or controls, and for the matching decision that put the load on the AV in the first place. A carrier that dispatches an autonomous load into a weather event outside the ODD because the dispatcher did not check the conditions has not transferred its liability to the AV vendor; it has created a coverage gap by operating outside the terms of the technology's certification.

Component Five: Customer Disclosure and Service Agreements

Not every shipper is comfortable with autonomous freight on their loads in 2026. Some are enthusiastic. Some have internal policies that require disclosure. Some have contract language requiring human drivers that predates the autonomous wave and has not been updated. Before a carrier books autonomous capacity against a customer's freight, the carrier needs to know which category each customer falls into.

The practical approach is a brief customer disclosure process: notify the shipper that the carrier now has autonomous capacity available on certain lanes, describe what that means operationally (the load moves on an AV within a certified corridor, there is remote monitoring, there is a defined exception protocol), and ask for a written acknowledgment if the customer requires it or a written objection if the customer does not consent. This process also protects the carrier: a customer who later objects to autonomous delivery cannot claim they were not informed if the carrier has a disclosure on file.

For contract customers, a rate agreement addendum is worth the administrative effort. Autonomous capacity may price differently than human-driven capacity on certain lanes (the cost structure is different when the AV vendor absorbs the driver expense), and a rate agreement that specifies the autonomous pricing separately protects both parties from ambiguity about what was quoted and what was delivered.

Component Six: Performance Measurement and Continuous Improvement

The mixed fleet is not a steady state. It is a transition, and the carrier's operating plan needs a measurement framework that tracks whether the autonomous operations are performing as intended and informs decisions about which lanes to expand, which to pull back from, and what the human-operations comparison looks like. The KPIs (key performance indicators) for the autonomous segment should include on-time delivery rate versus the carrier's human-driven benchmark for the same lanes, cost per loaded mile versus human-driven equivalent, exception rate (the frequency with which an autonomous load triggers a safety hold, a route suspension, or a hub handoff failure), and customer satisfaction scores for autonomous versus human-delivered loads.

The comparison discipline matters because it catches the hidden costs that vendor demos never show. An autonomous lane that delivers on time 97 percent of the time sounds excellent until you discover that the 3 percent exception rate requires expensive emergency human-driver insertions that wipe out the per-mile cost savings. The measurement framework catches that math before the carrier has scaled the autonomous lanes to the point where the hidden cost is a structural problem rather than an edge case.

Integrating Autonomous and Human Dispatch on One Board

One of the most common mistakes carriers make in early mixed-fleet deployments is treating the autonomous loads as a separate program managed by a separate person. The autonomous loads are freight. They need to be on the dispatch board with everything else, visible to the dispatchers who manage the network, and subject to the same load-planning discipline as human-driven loads. Keeping the autonomous loads in a separate system or a separate workflow creates visibility gaps, makes it harder to optimize the network as a whole, and often results in the autonomous capacity being underutilized because the dispatchers who see the load board every day are not seeing the autonomous slots.

The McLeod TMS integration addresses this by surfacing Aurora capacity in the load board alongside human-driver capacity. The dispatcher sees available capacity, not a technology category. The TMS configuration determines whether the AV slot is flagged visually (it typically is, for the confirmation step described earlier) and what information the dispatcher sees about ODD eligibility before committing. But the load-planning logic that decides which loads go to which capacity type is still the dispatcher's judgment, informed by the TMS's eligibility filters.

The dispatching rules for a well-configured mixed fleet should include: priority logic (for example, autonomous capacity is offered first for full-autonomous lanes to maximize asset utilization, human drivers are offered first for lanes with urban last-mile complexity), HOS-aware sequencing (the autonomous vehicle does not have an HOS clock, which means it can take the load that would require a human driver to use a restart; the TMS should surface this advantage automatically), and surge protection (a protocol for what happens during a weather event that removes a large portion of autonomous capacity from ODD eligibility, requiring rapid reallocation to human drivers).

The HOS point deserves a deeper look. When an autonomous vehicle completes a long corridor run, it arrives at the destination hub available for immediate reload. There is no 10-hour off-duty requirement, no 34-hour restart, no split-sleeper calculation. The AV availability profile is fundamentally different from a human driver's, and the TMS should be configured to leverage that difference rather than treating AV capacity as if it were just another truck with an invisible driver. A carrier that understands this dynamic and configures its TMS accordingly will be able to move more freight with the same combined asset base than a carrier that uses AV slots like slow-turnaround human loads.

The Compliance Layer: FMCSA, Insurance, and the Audit Trail

FMCSA is updating its HOS regulations to address driverless operations, a recognition that the existing rules were written for human drivers and do not map cleanly onto a vehicle that has no driver to accumulate duty time. The carrier planning for a mixed fleet needs to follow these regulatory updates closely, because the compliance obligations for autonomous loads under the updated rules are not simply "no HOS rules apply." The FMCSA is working through questions about vehicle inspection requirements, break requirements at transfer hubs, and the documentation obligations for carriers operating vehicles without a human driver aboard.

The CSA (Compliance, Safety, Accountability) scoring system, which FMCSA uses to track carrier safety performance and trigger interventions, was also designed for human-driven operations. A carrier that has an autonomous-related incident needs to understand how it will be scored under the CSA framework, because the existing categories (unsafe driving, hours-of-service compliance, vehicle maintenance) do not map cleanly onto an autonomous event. A carrier caught off guard by an adverse CSA score from an autonomous incident is a carrier that did not build the compliance question into its mixed-fleet operating plan.

The audit trail for autonomous operations should be as complete as the audit trail for human-driven operations, and in some respects it needs to be more complete because the accountability chain is less obvious. For a human-driven load, the ELD records the driver's duty status, the load tender records the dispatch, and the POD (proof of delivery) records the delivery. For an autonomous load, the equivalent trail includes: the TMS booking record (who committed the autonomous assignment and confirmed ODD eligibility), the AV vendor's route log (what the system did, where it deviated, any safety events triggered), the hub handoff record (what human took possession and completed the DVIR), the delivery confirmation, and any exception records. That trail needs to exist, needs to be stored in a format the carrier can retrieve, and needs to be linked to the load record in the TMS.

Carriers should assume that FMCSA and commercial insurers will want to see this audit trail in the event of any incident or claim, and that the carrier who cannot produce it will be in a worse position than the carrier who has it organized and accessible. Building the trail into the standard operating procedure from the first autonomous load is far less expensive than reconstructing it retroactively after a claim.

Key Takeaways

• Autonomous capacity is bookable today: Aurora's 250,000-plus driverless miles, accessible through the McLeod TMS integration serving 1,200-plus fleets, means the mixed fleet is not a future planning exercise but a current operating reality that requires a written operating plan before the first load moves.
• The ODD (operational design domain) defines where an SAE Level 4 autonomous vehicle can operate without human backup. Lane selection for autonomous deployment must begin with mapping existing lanes against the published ODD, producing three categories: full-autonomous lanes (complete corridor within ODD), hub-and-spoke lanes (AV handles the long middle leg, human drivers handle first and last mile), and human-only lanes (outside ODD or practically unsuitable for autonomy in the current cycle).
• A complete mixed-fleet operating plan requires six components: TMS dispatch routing rules with ODD eligibility flags and a human confirmation gate; safety and exception protocols for remote monitoring and AV suspension scenarios; transfer hub and handoff logistics including DVIR responsibility at the relay point; insurance and liability structure confirmed in writing with the carrier's broker before the first load; customer disclosure and service agreement addenda for rate and consent clarity; and a performance measurement framework that tracks autonomous versus human-driven KPIs on matched lanes.
• AV capacity should be managed on the same TMS dispatch board as human-driver capacity, not in a separate workflow. The McLeod-Aurora integration surfaces autonomous slots alongside human loads; the carrier's dispatch routing rules determine the priority logic, but visibility to all dispatchers is essential for network optimization.
• The HOS (hours-of-service) advantage of AV capacity is structural: an autonomous vehicle has no duty time accumulation, no 10-hour off-duty requirement, and no 34-hour restart. A TMS configured to leverage this availability profile will move more freight per asset than one treating AV capacity like a human truck with an invisible driver.
• FMCSA is updating HOS and inspection rules for driverless operations. The CSA (Compliance, Safety, Accountability) scoring framework was designed for human-driven fleets. Carriers operating autonomous loads need to monitor regulatory updates actively and build an autonomous-load audit trail that parallels the ELD and POD records used for human operations.
• The autonomous transition is a gradual shift, not a replacement event. Planning the mixed fleet well means more freight capacity from the same network while keeping human drivers in the roles, including first and last mile, transfer hub operations, and exception response, where their judgment is irreplaceable.
• Performance measurement is the discipline that prevents silent cost overruns. Autonomous lanes that show excellent on-time rates may still have exception costs (emergency human insertions, hub failures, weather-related diversions) that erase the per-mile savings. Track exception rates and total-cost comparisons, not just on-time rates, before scaling autonomous capacity.