Park Assist Optimization by Sensor Fusion Strategy: Development and Model-in-the-loop Validation

The Rear Park Assist feature is an ADAS (Advanced Driver Assistant System) function intended to provide warnings to the driver when an obstacle that can collide with the rear bumper of the vehicle is present near the back of a car moving in reverse. A Value Optimization project has been completed, with the support of Maserati S.p.A., to study the feasibility of a reconfiguration of the Rear Park Assist feature. Indeed, on current vehicles on the market, the Rear Park Assist function is based on ultrasonic sensors; but the availability of a rear-view camera and corner radars on some Maserati vehicle models may permit the feature to work as planned while relying on a sensor fusion strategy, which combines data coming from the two sensor kinds – camera and radars – and introduces several benefits following the elimination of ultrasonic sensors. To achieve the above goal and benefit from the foreseen advantages, a design stage, subsequent to a preliminary feasibility analysis for the project, has been carried out to produce algorithms for computer vision, radar targets elaboration, and sensor fusion which allow the detection of obstacles as intended by the feature. The validation phase, succeeding the development stage, has permitted to test the devised system against the expected performance documented in a related validation standard. In particular, the developed feature satisfied most of the requirements, although some undesired results mainly related to wrongly detected obstacles. An enhancement process, accompanied by on-vehicle calibrations and tests, might finally allow the Rear Park Assist system based on rear-view camera and corner radars to be implemented on series vehicles.