Assessing Autonomous Mobility-on-Demand Services and the Impacts of Operational Strategies: A Case Study of Chengdu, China

TEXT START: The Autonomous Mobility-on-Demand (AMoD) service is emerging as a potential alternative to on-demand urban mobility, but its operational performance relative to traditional street-hailing services and the effectiveness of related operational strategies remain unclear.

The Dissection

This is a narrow coordination-efficiency study disguised by the larger autonomous-mobility frame. It compares decentralized street-hailing with centrally dispatched vehicles under fixed historical demand, fleet constraints, and road networks. Its reported reductions—73.3–83.4% in passenger waiting time, 75.0% in deadheading mileage, and 74.0% in deadheading energy consumption—show how much friction centralized control can remove inside the simulation.

The important result is not merely faster rides. Human street-hailing is revealed as a wasteful matching layer. Repositioning, geofencing, request rejection, and fleet management become executable control functions rather than improvised driver behavior. If autonomous execution is real, the driver stops being a coordination asset and becomes an operating cost.

The paper does not model the actual transition: ownership, vehicle capital costs, maintenance, charging, safety intervention, regulation, pricing, congestion, revenue, or displaced labor. It optimizes vehicle circulation, not a viable post-labor industry.

The Core Fallacy

The central inferential error is treating operational superiority as economic replacement. The paper itself uses cautious language—“simulated,” “estimated,” and “potential”—so this is primarily an omission and a likely reader overreach rather than an explicit false claim.

Under the Discontinuity Thesis, the study provides a narrow signal for P2: centralized machine coordination can outperform human-mediated coordination. It does not establish P1 across cognitive work, nor does it prove P3—the collapse of productive participation. A fleet that moves passengers efficiently may still be uneconomic, politically blocked, unsafe, or dependent on a large invisible human support apparatus.

Hidden Assumptions

• Autonomous vehicles can execute the simulated service reliably without substantial human supervision or intervention.
• Historical demand remains unchanged despite lower waiting times, altered prices, different availability, and possible induced demand.
• Equal fleet-size constraints create a meaningful comparison while excluding capital, maintenance, insurance, charging, and labor costs.
• The graph-based matching mechanism adequately represents real-time traffic, vehicle failures, passenger behavior, accessibility needs, and exceptional trips.
• APWT, ADM, and ADEC are sufficient proxies for service quality and system viability.
• Repositioning, geofencing, and request rejection can be legally and operationally imposed at scale.
• The pronounced gains in airports, remote areas, and low-demand hours are representative of durable system-wide advantage rather than favorable simulation conditions.
• Energy consumed while driving is a sufficient energy metric; charging infrastructure, battery degradation, and upstream energy costs are absent.
• The fleet’s owner captures the coordination gains rather than drivers, platforms, regulators, or infrastructure providers absorbing them through competition and compliance costs.

Social Function

Classification: partial truth, transition management, and prestige signaling.

The paper contains a real and strategically important result: centralized dispatch can annihilate large quantities of matching waste. But it translates a potential labor-displacement event into safe technical metrics—waiting time, mileage, and energy consumption. That vocabulary allows institutions to discuss automation without stating the blunt consequence: once the fleet is autonomous and centrally controlled, human street-hailing labor becomes a cost center competing against software and capital.

It is not pure copium. The numbers point toward a genuine coordination advantage. Its anesthetic function lies in leaving ownership, surplus capture, and the fate of displaced workers outside the frame.

The Verdict

Technically, this is evidence that centralized autonomous mobility can outperform human street-hailing under controlled conditions. Systemically, it is not proof that robotaxis are already deployable or profitable; it is evidence of the mechanism that could kill the existing model. The paper supports a domain-specific P2 signal and points toward P3, while leaving the decisive question unanswered: who controls the vehicles, energy, logistics, maintenance, and access after the human coordination layer is removed?