Head-up display (HUD) systems reduce driver visual fatigue and improve driving safety. Existing HUD only alleviates visual fatigue by reducing the times a driver looks at the dashboard. However, conventional augmented reality heads-up display (AR-HUD) system can usually only image at a fixed projection distance. When drivers approach an object, it is perceived that the image passes through the object. This causes drivers to constantly distinguish the distance between the object and the virtual image, which will lead to visual fatigue. Indeed, there is a need for a new type of HUD that makes the projection distance variable. Existing designs use an off-axis three-mirror system, by adjusting the position of the first mirror to achieve the projection distance variable. However, it will change the down angle and also need the picture generation unit (PGU) with variable angle of light output. This will increase the cost and design difficulty. It is necessary to study another way to realize the function of variable projection distance.

Unlike traditional co-axial zoom optics, for HUD with variable projection distances, changing focus is not the only way. Changing the object distance and size can also realize it. We use changing magnification to realize variable projection distance and only change the position and size of the PGU. This is because, like looking in a mirror, as the object distance is larger, the virtual image is farther away. Furthermore, the distance between the PGU and the first mirror decreases at the same time as the projection distance decreases. Since none of the images of PGU emitting by the way of parallel light, as the PGU is closer to the reflector, the image reflected by the mirror is smaller. As the projection distance decreases, the tensor angle of the virtual image relative to the human eyes decreases. This design does not cause the down angle to change, and there is no need to use the PGU with a variable angle of light output. The off-axis three-mirror system generally determines the initial structure by calculating the curvature and optical spacing of the co-axial tri-reflector system. Then we adjust the off-axis angle and optimize the free-form mirror. It is complicated to obtain the different position parameters of the PGU through theoretical calculations. The theoretical data also need to be modified in combination with simulation, which will increase the workload. We design“macro”to obtain the relevant parameters directly through simulation to improve efficiency. The optimization is based on the projection distance of 10 m to optimize the free-form surface and determine the off-axis angle. The change step of the projection distance is 50 mm, and the change step of the image source size is 0.01 mm. We use random sampling to verify the function of continuously variable projection distance of the virtual image. The image quality evaluation includes transverse vertical (TV) distortion, rhombic distortion, trapezoidal distortion, grid distortion, dynamic distortion, binocular parallax, image tilt, aspect ratio distortion, MTF, spot diagrams, and field of view distortion. We evaluate the image quality in detail to better assess the imaging situation after actual manufacturing and guide suppliers to improve their products. The above information is represented in the tables and simulation diagrams.

We use“macro”to obtain parameters related to the continuous variation of the projection distance. The image quality is analyzed when the projection distance is 10 m and 3 m respectively. The image quality is analyzed by random sampling to randomly select any projection distance within the changing range. The RMS radii of the spot diagrams in the three cases are all within the Airy spot and less than 25 μm. In addition, the grid distortion, TV distortion, rhombic distortion, and trapezoidal distortion are all less than 5%. Dynamic distortion and binocular parallax are less than 5'. The field of view distortion and aspect ratio distortion are less than 5%. Mutually using parameters of a projection distance apart 50 mm does not result in a dramatic change in image quality. In the range of the variable projection distance, the image quality meets design requirements. The designed projection distance can be continuously varied and the change step of the projection distance setting is reasonable.

The AR-HUD with variable projection distance can further reduce the driver's visual fatigue and improve driving safety. The means of equipment and software correction can affect the criteria for assessing image quality, but the design direction is the same. The design value of each distortion parameter should be as small as possible when conditions allow. This can give the actual manufacturing enough tolerance range. Within the space constraints of the car, we need to trade off dynamic distortion and mesh distortion. Under limited conditions, it is not possible to consider both distortions at the same time. In general, the focus should be on controlling dynamic distortion rather than mesh distortion. Appropriate distortion can be corrected by software. Compared to several design approaches, this design does not pose the risk of down angle perspective changes. The method of this design is applicable to all kinds of HUD systems based on the off-axis three-mirror system. In addition, we give the optical detection methods for reference. While our HUD architecture is relevant, the method of realizing variable projection distances is general in nature.