The mission earlier this year didn't succeed because of communications problems. It looks like there's not going to be any available data on performance of the system.
The intended sequence starts when
the approach phase is initiated and Jerry begins to automatically transmit its absolute orbital data to Tom via the inter-satellite link system. Tom maneuvers closer to Jerry based on the obtained relative orbit estimation. During the mission design stage, it was initially assumed that a single electrical thruster could be used to position Tom. However, it was subsequently determined that the thrust of the propulsion system would not sufficiently satisfy the expected performance requirements: whereas the theoretically determined thrust of the propulsion system was 200 μN, only 50 μN turned out to be available following development. Therefore, an alternative orbital control method using air drag was proposed. Tom controls its attitude to maximize and minimize air drag according to the along-track position secular terms given under Hill’s dynamics, while the originally designed propulsion system is used intermittently upon request by the ground station.
Validation of the vision alignment system during the mission phase is initiated via telecommand. Jerry orients the Sun and activates its LED markers while at the same time Tom controls its relative orbit to maintain its position with respect to the Sun and Jerry. Tom determines whether Jerry is in the field of view of the visible-light camera and then estimates the relative position and attitude, which is used to perform feedback control to maintain the satellites’ alignment for a few minutes. In this simulation, the alignment is maintained for about 2 min by utilizing the propulsion system to implement a maximum thrust level of 50 μN. After the two CubeSats have demonstrated the operation of the vision alignment system, their mutual distance increases. Tom then repeats its approach to Jerry, and the process is repeated. In this manner, the vision alignment system is re-validated over the mission phase.
The "Vision Alignment System" is described in the construction paper as:
a vision alignment system that can determine the relative position and relative attitude between the two CubeSats at the same time. The vision alignment system consists of laser beacons on Jerry and one visual camera loaded on Tom. Tom detects the laser diode projection images to determine the relative position and relative attitude between the two CubeSats. In vision alignment mode, Jerry will fix its attitude for the laser diodes on Jerry such that they look at the Sun. Then, Tom will interrupt this line between Jerry and the Sun and will keep the inertial alignment between Jerry, Tom and the Sun during a few minutes. To do this, Tom will fix its attitude with respect to the Sun and will control its position on the transversal plane of Jerry and the Sun alignment by using reaction wheels and thrusters (Figure 1).
(from Figure 1 of the construction paper)
It's using vision, in the sense of imaging, to recognize and measure the (projected) position of the laser sources, and from that calculating alignment and distance.
The planned performance wasn't all that aggressive:
The goal was to control position to better than a meter, and measure that position to better than 10cm.
How does that connect to the needs of a "virtual telescope"? I don't think they're talking about the wavelength-scale alignment needed by e.g. the segments of a composite mirror. Rather, they seem to be focusing on (pun intended) the larger structural elements of a Really Big telescope:
Current space telescopes have a single structure, and consequently, their focal length cannot be elongated sufficiently. Sometimes, this problem prevents the improvement of the resolution of space telescopes. In order to resolve this problem, the concept of Virtual Telescope is proposed. A Virtual Telescope consists of two spacecraft; one has a lens system, and another has a detector system. By using formation flying, they can be simplified as a virtual telescope system. Then, the relative orbit distance of two spacecraft can be a baseline of a virtual telescope system (1,2). The most important point of Virtual Telescope is to build the inertial alignment with respect to a celestial object and to maintain it in space environment. Inertial alignment means that relative position and relative attitude of the two spacecraft are simultaneously aligned with a target.
The first reference there is "THE VIRTUAL TELESCOPE DEMONSTRATION MISSION (VTDM)" N Shah et al 5th International Conference on Spacecraft Formation Flying Missions and Technologies, May 29-31, 2013, which talks about a 100m separation demonstration for "a “virtual telescope” (VT) with very long focal lengths (20 m to >1,000 km)". They proposed a mission with more stringent alignment goals than CANYVAL-X, but not all that more stringent: Still millimeter scale.
The second reference is “Orbit Design and Control of Technology Validation Mission for Refractive Space Telescope in Formation Flying,” AIAA Guidance, Navigation, and Control Conference, SicTech 2014, January 13 - 17, 2014 with overlapping author list; it's paywalled but basically talks about the same kind of thing. To me, the most interesting part of that reference is the discussion of a refractive space telescope. That's perhaps a topic for another day, but it's part of the motivation for the long focal length.
Coming back to the main point of the Question, the goal of these projects seems to be to create mm-scale alignment across 100m- or km-scale free-flying structures. They would then use active optical elements in the telescope system to get down to their final precision. That's still an advancement in station-keeping, but not nearly as intense an advance as it first seemed.