The magic of this hypothetical robot is confined to its “brain,” which is supposed to be a full-fledged quantum computer (QC). The rest of the robot is an extended classical Braitenberg vehicle, that is, two rear wheels and a propeller (that’s new), each connected to an independently operating motor. Two light sensors in the front detect incoming light on the left and right side of the symmetry axis of the vehicle, respectively.
Following Braitenberg, the vehicle will engage only the right rear wheel when sensing light only with its right sensor, thus making the vehicle do a left turn away from the light source (the same with sides flipped). In case no sensor detects light, the vehicle will simply move forward, powering both rear wheels. If both sensors detect light (beyond Braitenberg’s original 1986 model), the vehicle will engage solely the propeller, thus lifting it up, like “jumping” over an obstacle.
At this point, you might rightfully wonder where in this wholly deterministic setting the “fun” part of QC would set in, such as entanglement, exponentially fast algorithms, or probabilistic measurements. The lackluster answer is that the authors, not unoriginally but uninspiringly, identify a quantum algorithm using two CNOT and five Toffoli gates working on a five-dimensional Hilbert state, exactly reproducing this entirely causal behavior. This is then confirmed by a simulation on IBM Quantum Experience using 78 even more basic gates (like Hadamard, S, or T gates).
A final rewriting of this robot’s behavior in terms of a highly artificial “game” of avoiding obstacles does not add any new insight.
Unless you are really interested in how a handful of trivial “if” statements on two input bits may be exactly reproduced in a (complicated) quantum circuit, you might safely ignore this paper; it does not contain any particularly interesting robotics or QC-related content.