Electromagnetic wave interaction mechanism and constraints of physical dimensions
The primary reason for the sparseness of millimeter wave radar point clouds stems from the basic physical laws of wave optics and electromagnetism. The mainstream working frequency band of vehicle-mounted millimeter wave radar is 77GHz to 79GHz, and the corresponding wavelength is about 3.8mm to 3.9mm.
According to the electromagnetic wave reflection theory, the relative roughness of the object surface determines the characteristics of the echo. When the detection wavelength is much larger than the undulation size of the object surface, the surface appears as a quasi-mirror surface from the perspective of electromagnetic waves, and the resulting reflection follows Snell's law, that is, the incident angle is equal to the reflection angle.
In urban road scenes, the metal surfaces of cars, glass curtain walls of buildings, and flat asphalt pavements are almost all "mirror surfaces" for millimeter waves with wavelengths close to 4mm.
This specular reflection causes most of the electromagnetic energy to dissipate in a direction away from the millimeter-wave radar, with only a very small amount of energy being transmitted back to the receiving antenna through diffraction at the edge of the object, secondary reflection from the corner reflector structure, or backscattering from normal incidence.
In contrast, the wavelength used by lidar is at the 905nm or 1550nm level, which is three orders of magnitude smaller than millimeter waves. Many object surfaces are rough for lasers and can produce uniform diffuse reflection, thus ensuring that all parts of the object surface can reflect echo points.
In addition to differences in reflection patterns, the dielectric constant and conductivity of the material itself also affect the richness of the point cloud. As a good conductor, metal has extremely high reflectivity for millimeter waves, so vehicles, guardrails and other objects can form relatively stable detection points. For non-metallic targets such as pedestrians whose main component is moisture, the absorption and scattering mechanism of millimeter waves is more complex.
Although the carbon content of the human body makes it somewhat reflective in the millimeter wave band, because the surface shape of the human body is extremely irregular and does not have a large area of planar or angular reflection structure, the energy is easily scattered in multiple directions, causing the echo intensity to fluctuate violently.
Some studies have done experiments on this. The use of carbon-coated human body models can simulate the reflection characteristics of pedestrians. However, even so, when the pedestrian's limbs are at an angle relative to the radar ray, a large number of radio frequency signals will be deflected instead of returned. This also explains why in the millimeter-wave radar view, the point cloud of pedestrians is not only sparse but also often missing parts.
The limitations of hardware aperture and angular resolution further exacerbate the discretization of spatial perception. The ability of millimeter wave radar to distinguish adjacent targets is limited by the angular resolution of the antenna, which is physically determined by the ratio of wavelength to the equivalent aperture of the antenna.
Limited by the vehicle installation space, the physical size of millimeter wave radar antennas cannot be expanded infinitely. This makes the horizontal angular resolution of traditional millimeter wave radars only maintain between 5° and 10°, and most of them do not have the ability to perceive pitch angles.
This means that within a wide beam range, even if there are multiple reflection centers, millimeter wave radar may merge them into a single point output due to insufficient resolution. This inefficiency at the "spatial sampling" level fundamentally limits the number of point clouds that can be generated in a unit space, making it impossible for millimeter-wave radar to build detailed three-dimensional models through dense laser beam scanning like lidar.
