On June 26, 2026, the National Highway Traffic Safety Administration (NHTSA) published a Notice of Proposed Rulemaking (NPRM) that will dismantle a 31-year-old assumption embedded in federal safety compliance: that every light vehicle on US roads has a physical brake pedal awaiting a human foot. The landmark proposal explicitly amends Federal Motor Vehicle Safety Standard (FMVSS) No. 135 (Light Vehicle Brake Systems) to establish distinct requirements for vehicles that use automated driving systems (ADS) without any manually operated controls.
For autonomous platforms built without a steering wheel or traditional foot controls, the mandate for a physical cabin interface disappears. Crucially, however, the physical stopping distances that vehicle dynamics and computer-aided engineering (CAE) teams validate against remain entirely unaltered. For testing departments, the challenge is structural: while the physical targets haven’t moved, the actuation command, data instrumentation and type-approval certification package must be completely reengineered.
Bridging the legacy gap
First issued in 1995, FMVSS No. 135 defines stopping distance as the exact interval between the moment force is applied to the physical brake control, to the point at which the vehicle comes to a complete stop. The historic framework relies on that force originating from a foot-operated control. Although this works seamlessly for driver-operated setups, it introduces severe engineering anomalies for ADS platforms where braking commands manifest as electronic controller area network (CAN) packets routed to a linear actuator.
Furthermore, NHTSA acknowledges that for fully autonomous robotaxis, a physical pedal assembly represents an active cabin safety hazard. An occupant accidentally or intentionally depressing a legacy pedal during automated operations could induce severe systemic conflicts, overriding or corrupting the safety logic of the automated driving system.
Evolving definitions of actuation
The new NPRM restricts traditional foot-pedal and manual parking brake rules strictly to vehicles maintaining physical controls. For purpose-built ADS vehicles, the service and parking brakes must be triggered exclusively by onboard computing architectures, with external remote command signals strictly precluded.
To bridge the engineering gap, the proposal introduces a broader definition of a ‘service brake control.’ For pedal-less designs, this includes any physical or solid-state component that translates an incoming electronic command into mechanical input for the hydraulic or brake-by-wire system. Track testing and laboratory procedures will be modified accordingly: standard pedal-force requirements will be set aside in favor of manufacturer-provided specifications defining baseline control inputs. Additionally, the traditional dashboard warning telltale must be engineered to be clearly visible from any designated seating position inside the cabin.
The physical limits of braking performance remain untouched. A pedal-less robotaxi must continue to demonstrate a cold-effective stopping distance of 70m or less from an initial velocity of 100km/h. Thermal fade requirements remain equally rigid, demanding that at least one of two consecutive hot stops from 100km/h registers at 89m or less. Wheel lockup constraints remain capped at 0.1 seconds above speeds of 15km/h, and manual anti-lock braking system (ABS) override controls remain strictly banned.
The data instrumentation challenge
While the physical performance dimensions of the NPRM are unambiguous, establishing compliant verification procedures introduces serious friction for track-test engineers. Legacy testing protocols rely heavily on a calibrated load cell mounted directly to a physical pedal to measure normalized occupant force. In an ADS vehicle, that physical force is replaced by a decentralized network command, be it a target deceleration rate, a specific torque request sent directly to a brake-by-wire module, or a positional target for a master-cylinder actuator.
Currently, there is no standardized industry consensus identifying which digital bus signal represents the definitive equivalent of a human ‘panic stop’ application. In response, NHTSA has opted to defer to individual OEM specifications. While this approach maintains strict technology neutrality, it shifts a substantial engineering burden onto the regulatory certification package. Homologation teams must clearly define a ‘maximum input’ condition that aggressively stresses the hydraulic lines, document all underlying signal architectures, and satisfy compliance auditors that the chosen software state genuinely simulates a worst-case emergency stop. Consequently, third-party compliance testing facilities will require comprehensive access to proprietary signal dictionaries to inject these explicit commands directly onto the vehicle network bus alongside existing deceleration and wheel-speed data acquisition (DAQ) channels.
The thermodynamic compensation conundrum
This operational paradigm shifts the focus toward brake fade and thermal recovery tests. Traditional testing stresses thermal margins through predictable, repeated manual pedal applications. Conversely, an autonomous vehicle governed by an integrated deceleration controller will dynamically adapt to fading friction coefficients by automatically ramping up its internal line pressure or torque requests to hit its deceleration targets.
This behavior forces a technical impasse that the industry must resolve: should the controller’s active thermal compensation logic be considered a native component of the system under evaluation, or must that closed-loop logic be inhibited during compliance validation to isolate and characterize the raw, underlying hardware? The current NPRM leaves this operational parameter unresolved, creating a vital target for industry feedback.
Strategic action before the deadline
The proposal explicitly sidesteps how passengers might interact with an emergency-stop sequence or how the ADS core should arbitrate such inputs. For the immediate future, any unforeseen passenger-interaction risks will fall under NHTSA’s broad defect enforcement authority, while a comprehensive, dedicated ADS performance standard is drafted for long-term implementation. Ultimately, the revised FMVSS No. 135 framework validates whether an autonomous vehicle is mechanically capable of stopping within a safe physical boundary once a command propagates; it does not evaluate whether the perception stack or path-planning software calculated that command at the appropriate timestamp.
Formal industry comments under docket NHTSA-2026-0728 must be submitted by July 27, 2026. This proposal marks the fifth standard modernized under the AV Framework, yet it represents the absolute first iteration where hardware removal alters real-world physical performance measurements rather than secondary cabin convenience features. For engineering units inside OEMs, Tier 1 system suppliers and testing laboratories, focus must remain on providing highly granular feedback regarding signal documentation baselines, closed-loop compensation parameters during thermal degradation, and standardizing passenger-system interactions. The physical removal of the pedal makes for an easy headline, but the underlying narrative is a complex, line-by-line adaptation of a 1995 protocol to an entirely digital machine.
Sources
• NHTSA Notice of Proposed Rulemaking, “Federal Motor Vehicle Safety Standards; Modernization of FMVSS No. 135 To Accommodate ADS-Equipped Vehicles,” Federal Register, 26 June 2026 (Docket NHTSA-2026-0728).
• NHTSA press release, “Trump’s Transportation Department Launches Commonsense Updates to Brake Pedal Requirements for AVs,” 25 June 2026.
• 49 CFR § 571.135, “Standard No. 135; Light vehicle brake systems.”
• Sidley Environmental, Health, and Safety Brief, “NHTSA Proposes Amending Federal Brake Standards for Autonomous Vehicles,” 29 June 2026.
