System and method for generating a point cloud based upon radar data
AssigneeGM Cruise Holdings LLC
InventorsMichael Meyer, Marc Unzueta Canals, Kalin Hristov Kabakchiev, Jessica Bartholdy Sanson
Life of this patent
Estimating the term…
Abstract
The technologies described herein relate to computing a point cloud based upon data output by several radar sensors in a distributed radar system. The radar sensors generate tensors based upon echo signals detected by the radar sensors. Values are extracted from the tensors and a sequence of tokens is created, where the sequence of tokens includes the values extracted from the tensors. The sequence of tokens is provided as input to a transformer model, which outputs a point cloud based upon the sequence of tokens.
Claims20
A computing system comprising:
- a processor;
- and
- memory storing instructions that, when executed by the processor, cause the processor to perform acts comprising:
- obtaining first values from a first tensor generated by a first radar sensor, where the first values are extracted from first tensor bins of the first tensor that correspond to a distance range;
- generating a first sequence of tokens based upon the first values;
- obtaining second values from a second tensor generated by a second radar sensor, where the second values are extracted from second tensor bins of the second tensor that correspond to the distance range, the first values and the second values temporally corresponding to one another;
- generating a second sequence of tokens based upon the second values;
- providing the first and second sequences of tokens as input to a transformer model, where the transformer model generates a point cloud for the distance range based upon the first sequence of tokens and the second sequence of tokens;
- and
- controlling an operation of a mechanical system based upon the point cloud generated by the transformer model.
A method performed by a computing system that is in communication with a distributed radar system, the method comprising:
- obtaining first values from a first tensor generated by a first radar sensor in the distributed radar system, where the first values are extracted from first tensor bins of the first tensor that correspond to a distance range;
- generating a first sequence of tokens based upon the first values;
- obtaining second values from a second tensor generated by a second radar sensor of the distributed radar system, where the second values are extracted from second tensor bins of the second tensor that correspond to the distance range, the first values and the second values temporally corresponding to one another;
- generating a second sequence of tokens based upon the second values;
- providing the first and second sequences of tokens as input to a transformer model, where the transformer model generates a point cloud for the distance range based upon the first sequence of tokens and the second sequence of tokens;
- and
- controlling an operation of a mechanical system based upon the point cloud generated by the transformer model.
A platform comprising:
- a mechanical system;
- a distributed radar system, the distributed radar system comprising a first radar sensor with a first antenna and a second radar sensor with a second antenna;
- and
- a computing system that is in communication with the radar system and the mechanical system, wherein the computing system is configured to perform acts comprising:
- obtaining first values from a first tensor generated by the first radar sensor, where the first values are extracted from first tensor bins of the first tensor that correspond to a distance range;
- generating a first sequence of tokens based upon the first values;
- obtaining second values from a second tensor generated by the second radar sensor, where the second values are extracted from second tensor bins of the second tensor that correspond to the distance range, the first values and the second values temporally corresponding to one another;
- generating a second sequence of tokens based upon the second values;
- providing the first and second sequences of tokens as input to a transformer model, where the transformer model generates a point cloud for the distance range based upon the first sequence of tokens and the second sequence of tokens;
- and
- controlling an operation of the mechanical system based upon the point cloud generated by the transformer model.
Dependencies are inferred from the claim text.