Specify the job before comparing sensors
Before comparing LiDAR suppliers, specify the job. A system detecting an object at highway speed has a different time and distance budget from a robot approaching a shelf. The useful range, field of view, surface reflectivity and permitted failure behavior follow from that job. An architecture should be evaluated against those requirements before its headline specifications are compared.
LiDAR combines an optical measurement with packaging, timing, processing and integration. The source of differentiation may sit in the emitter, receiver, scanning arrangement, signal processing or manufacturing process. Those parts must work together at the customer's required volume and operating conditions. A single impressive specification does not identify the commercially decisive part.
Test the measurement under real conditions
Time-of-flight systems infer distance from the travel time of emitted and returned light. FMCW systems use frequency-modulated continuous-wave measurement and can provide a radial-velocity observation through Doppler information. Neither description establishes that every implementation performs equally well. Optical design, signal processing and operating conditions matter within each family.
Aeva's Atlas materials describe simultaneous range and velocity detection. That supports identifying the product's advertised architecture and capability. Comparative claims need a common test, including target properties and detection criteria. We do not convert the manufacturer's product description into an independent ranking of all LiDAR suppliers.
A simple timing calculation with explicit limits
For an illustrative target 100 meters away, light travels approximately 200 meters on the outward and return path. Using a rounded speed of light of 300 million meters per second gives a round-trip time of about 667 nanoseconds. A one-nanosecond timing difference corresponds to roughly 0.15 meters in inferred one-way distance.
This calculation describes a basic relationship, not a commercial sensor's resolution or accuracy. Real systems use signal processing, repeated observations and calibration, and have error sources beyond a clock interval. It would be incorrect to infer a product's measured performance directly from the simplified arithmetic.
Range should also be separated from object detection. Receiving a return does not establish that the downstream system can classify the object or respond safely. The 3D detection survey is relevant background for that distinction. The perception guide follows the broader chain from observation to action.
What belongs in a supplier comparison
Compare products at a stated target reflectivity, range, field of view, environmental condition and acceptable error rate. Include power, package size, thermal requirements, integration effort and the physical mounting constraints. Published maxima achieved under different conditions should not be assembled into an apparent like-for-like performance table.
Manufacturing readiness is another independent question. A working prototype does not establish yield, calibration throughput, field reliability or the ability to deliver accepted units repeatedly. The buyer's qualification and service process can create switching costs, but those costs must be demonstrated. They cannot be inferred from an optical patent or a large nominal market.
Design wins and revenue are separate milestones
A demonstration, development agreement, nomination, production order and recognized revenue are different events. A useful company update identifies which event occurred and links it to a dated disclosure. It also identifies contingencies, such as a customer's launch timing or qualification requirements, before treating the event as a commercial forecast.
This article names Aeva as a product example because its architecture is directly described in the linked primary material. It makes no current market-share ranking and does not infer the financial position of other vendors from historical product comparisons. Expanding the company table requires current company-specific evidence, not a list carried forward from an older research pass.
The decision that survives a changing market
The strongest supplier case combines a task-relevant measurement advantage with repeatable production and an acceptable ownership cost. The case weakens if another sensor arrangement satisfies the same requirement more cheaply, if the customer changes its system design, or if support and manufacturing costs consume the apparent differentiation.
For follow-up work, we would prioritize comparable field tests and dated production disclosures. A vendor specification is a starting point for diligence; the customer's accepted result is the commercial test. This guide therefore stands as a component-selection discussion, while the related perception article addresses system-level reliability. Neither is a forecast of universal LiDAR adoption.