You still only get one range and one velocity line per radar observation.
You can, however, "scan" a region (that's why airport towers and ships have these rotating antennas) and get another axis of information.
The same can be achieved using digital beamforming, which, however, requires multiple signal chains, and specific antenna arrays. For a two-dimensional picture, you'd need an at least two-dimensional array of antennas. For the resolution displayed, you'd need both high bandwidth (think in the order of 1GHz) as well as small wavelength (think of beyond 76 GHz). So, I'd say, without a millimeter-wave lab, this is probably not achievable.
I've not seen pictures like that myself (the article says "source: NXP" without elaborating any further, so this might just as well be "artist's interpretation of what NXP wants to achieve in 10 years).
Whilst this resolution might be impossible to achieve with commodity hardware, in principle, SDRs that are MIMO-capable do exist. If you can get your hand on a 4x4 MIMO phase-aligned SDR system, you could, behind/infront ambiguities notwithstanding, build a simple patch antenna array of 2 rows / 2 columns parallel patches and build a digital beamforming radar with that. You'd probably want to avoid the problems of isolating RX from TX by having a separate array for TX and RX.