ACAS X Archives - Leeham News and Analysis

ACAS X, the FAA-sanctioned next-generation collision avoidance system developed by MIT Lincoln Laboratory under federal contract, has been quietly transitioning from certification testing into operational cockpits while policymakers and aviation commentators debate the broader question of whether artificial intelligence belongs in aviation safety infrastructure. The Leeham News analysis by Vincent E. Bianco III, published May 24, 2026, draws a sharp distinction between that ongoing public debate and the regulatory reality already in place: the FAA approved an AI-augmented airborne collision avoidance architecture years before the current argument over Secretary Duffy's proposed SMART modernization framework reached its present volume. The article's central assertion—that the AI-and-ATC debate is three years too late—reflects a recurring pattern in aviation where front-line technology adoption outpaces the political and journalistic discourse surrounding it.

ACAS X represents a fundamental departure from the deterministic, rule-based logic that defined TCAS II, the system that has been mandatory on large transport-category aircraft since the early 1990s. Where TCAS II generates Resolution Advisories through a fixed decision tree encoded in ARINC 735B, ACAS X applies a probabilistic optimization framework rooted in dynamic programming and Markov Decision Processes—essentially a form of machine learning inference operating in real time against a continuous model of aircraft state and airspace geometry. The practical result is a system capable of issuing advisories that are geometrically smoother, less prone to nuisance alerts, and better suited to the high-density, mixed-equipage airspace that defines contemporary commercial and business aviation operations. For flight crews, ACAS X advisories are intended to be more actionable and less likely to demand the kind of aggressive vertical maneuvering that has caused secondary conflicts under TCAS II in congested terminal environments.

The article's framing around "SMART"—Transportation Secretary Sean Duffy's ATC modernization initiative, which has attracted both industry support and skepticism regarding its AI integration proposals—highlights a jurisdictional and philosophical divide that working pilots and operators should recognize. The regulatory architecture governing airborne collision avoidance falls squarely within FAA technical standards authority and has proceeded largely independent of the congressional and executive-branch debates over ground-based ATC automation. Operators flying under Part 121, 135, or business aviation Part 91K frameworks that already carry TCAS II-equipped aircraft will face eventual transition mandates to ACAS X as the system's deployment expands; understanding that ACAS X is not an experimental concept but a certified, fielded technology changes the practical planning horizon for avionics upgrade cycles, crew training curricula, and simulator fidelity requirements.

The broader trend this analysis reflects is the growing divergence between the pace of certified avionics development and the pace of public policy deliberation over AI in aviation. The FAA's Special Committee 147, which governs TCAS and ACAS standards through RTCA, has been working the ACAS X technical framework since the mid-2010s, and the system's probabilistic core has been scrutinized through the same DO-178C and DO-254 rigor applied to any safety-critical avionics function. That process—methodical, technically grounded, and largely invisible to the general aviation press—is precisely the regulatory architecture Bianco argues has already answered the foundational questions that the SMART debate is only now posing. For airline dispatchers, chief pilots, and avionics managers, the practical implication is straightforward: ACAS X is not a future consideration contingent on the outcome of policy debates in Washington. It is an active transition underway, and the organizations best positioned to manage it will be those tracking RTCA DO-385 compliance timelines rather than waiting for political consensus to catch up with the cockpit.