Page 18: of Marine Technology Magazine (Jul/Aug 2026)

RADAR PROCESSING

Detection, tracking, and obstacle reporting Obstacle and target data can be supplied to an external au-

Once a signi? cant return is identi? ed, adjacent signi? cant re- tonomy or vessel-control system using formats including AS- turns can be grouped together into a plot (detection). A plot is TERIX, NMEA-0183, TAK, and SAPIENT.

a single observation: something of interest appears to be pres- ent at a particular position during one radar scan. Combining sensors

Tracking adds history and trajectory. Tracking software com- A radar track may show that an object is approaching or on a pares plots between successive scans and decides which are likely steady bearing without identifying it. AIS may provide iden- to belong to the same object. From these linked observations, it es- tity data that can be correlated with the track, while a camera timates course, speed, and likely future position. Con? dence tests may offer visual con? rmation. Sensor fusion combines these help prevent unrelated detections from becoming a false track. observations into a more complete picture.

For small targets, tracking may become dif? cult as detection Successful fusion depends on consistent timing and coordi- strength and reliability reduce. This is where tracking from nate systems and an understanding of the accuracy of each video can help; a track that has already been acquired by tra- source, particularly where position is concerned. Bounding ditional detection and tracking methods can be maintained for boxes de? ning uncertainty margins are critical with the aim longer by tracking directly from the radar video. of maximizing the number of contact correlations while mini-

For autonomous navigation, it is not always appropriate to mizing false matches. Robust fusion is what turns multiple wait for a fully established track. Short-range or intermittent sensors into a coherent navigational picture.

detections of objects such as a ? oating container or a small A correlated track report links the individual sensor observa- mooring buoy may generate a hazard that needs to be reported tions that support it, allowing the receiving system to choose before a track can be generated. Proximity detections from a information from any of the inputs. For example, speed and radar processor report the closest returns meeting con? gured course can be taken from the radar track, object classi? cation criteria within a de? ned area around the vessel. and enhanced bearing measurement from a video track, and

Proximity detections are reported with very low latency, on cargo information from the AIS.

each scan, as the radar turns. Safety areas can re? ect speed, maneuverability, and environment. A forward zone might ex- Position integrity without GPS tend farther at higher speed, while a tighter zone may suit har- Autonomous vessels commonly depend on GNSS/GPS. bor maneuvers. Loss of signal is a problem, whereas jamming can prevent re-

Cambridge Pixel has been providing technology components liable reception and spoo? ng can cause a receiver to report a for USV and ASV developers for over a decade, including its plausible but false position.

SPx Server V2 software, which includes radar video process- In coastal areas, radar can provide an independent cross- ing, proximity detection and tracking for collision avoidance check of GNSS-derived positions. Cambridge Pixel’s GPS and autonomous navigation. Its Detection Server con? gura- Assist compares live radar data with a predicted radar image tion provides plot extraction and proximity detection, while generated from terrain and coastline information. By ? nding the Tracking Server adds multi-hypothesis target tracking. the alignment that best matches the two images, it estimates the vessel’s position independently of incoming GNSS data.

If the positions differ signi? - cantly, GPS Assist can raise an alarm and also generate an al- ternative NMEA-0183 stream to provide backup positional information.

Safe and effective maritime au- tonomy will not come from one sensor or algorithm. It will come from systems that provide inde- pendent, timely, and understand- able evidence. Radar is central to that picture, but raw echoes are not enough. Only through rigor- ous processing can radar deliver the detections, tracks and ob-

Cambridge Pixel’s SPx Fusion Server correlates track reports from stacle warnings that autonomy multiple radar sources and AlS into single fused tracks. systems depend on.

Credit: Cambridge Pixel 18 July/August 2026

MTR #5 (18-33).indd 18 MTR #5 (18-33).indd 18 8/5/2026 11:31:05 AM8/5/2026 11:31:05 AM