Radar is gradually emerging as the key technology for enabling intelligent and autonomous features in modern vehicles. From relieving the drivers from monotonous tasks to reducing stress and adding life-saving interventions, radar is proving to be the all-in-one technology for today's tech-powered vehicles. PathPartner’s automotive journey has witnessed several success stories, and it continues to work on various use-cases of modern-day vehicles like
DMS, In-cabin Sensing, etc. Santhana Raj, as a technical architect, has played a pivotal role in developing
radar-based ADAS at PathPartner. With over 12 years of experience in signal processing, he has published several research papers over the years. Some of the prominent ones include
Improved and Optimal DBSCAN for Embedded Applications Using High-Resolution Automotive Radar Bayesian Grouping of Multi Sensor Radar Fusion for effective Pedestrian Classification in Automotive Surround View Witnessing the demand for radar-based applications in automotive, PathPartner has developed a dedicated lab to make testing more accurate and closer to real-world scenarios. Let’s get to know Radar for Automotive from Santhana himself.
Q1: Can you give the readers a brief idea of PathPartner’s expertise with a particular focus on its Automotive offerings?
A1: PathPartner is a product engineering services company with diverse experience in BSP & device drivers, sensor integration, signal processing, computer vision, artificial intelligence, imaging, deep learning, machine learning, media codecs, and hardware design. With ADAS picking up pace in user adoption, we have been using our expertise in developing product accelerators and IPs to assist OEMs & Tier Is to accelerate their new product development and feature enhancement. We offer our accelerators across ADAS, IVI, and other vehicle electronics aimed at both safety functions and convenience features.
Q2: Can you tell us a bit about PathPartner Radar expertise?
A2: We have been working on our Radar SDK, which overcomes the need for initial development required by any tier 1 or OEM when using any radar chipset with varied antennae patterns. Our platform-agnostic SDK works with leading radar chipsets as well as custom radar boards. Radar SDK is a suite of algorithms consisting of multiple highly advanced algorithms in each block of radar raw data processing. Depending on the customer applications and the processing needs, we pick and choose one of the algorithms in each block and package it to meet the end application PIs for the customer.
Q3: What can you tell us about the various applications of radar in the Automotive industry?
A3: Radar is used in the exterior and interior of cars for applications such as ACC, AEB, BSD, and interior for applications such as unattended child monitoring or cabin monitoring. We have also been seeing unique applications such as curb height detection, door obstruction detection, auto boot opening, gesture recognition, and more.
Q4: What can you tell us about Radar feature development?
A4: Any development of a radar system involves following basic steps-
- Basic hardware design, including the RF antenna selection and design
- Choosing the right operating parameters, including chirp configuration and algorithm parameters
- Core algorithm development and implementation
- Performance optimization
Most prominently, calibration and testing play a significant role in Radar product development since algorithms have tunable parameters. It is critically important for one to do proper calibration followed by exhaustive testing. We at PathPartner have a dedicated lab to do our testing.
Q5: Can you explain more about the testing aspect for automotive radar applications?
A5: When I say lab for testing, automotive radar application is confined to replicating real-world obstacles that cars might face and not testing the RF performance. With little room to tune a few parameters, one can get their hands on a stock SDK straight from the radar board manufacturers. To verify this, one needs a dedicated lab with testing equipment that mimics real-world situations. We at PathPartner have equipment that mimics obstacles such as concrete walls, gates, metal walls, metal poles, curb, metal mesh, wooden blocks, grills, and more. Additionally, we also have absorber sheets and corner reflectors of various RCS capacities. On another spectrum, we also have dummies that mimic human heartbeat and breathing that helps in testing applications such as live object detection and occupancy monitoring. PathPartner with extensive experience in developing radar systems for automotive and industrial application was successful in providing algorithms and system optimization services to enable real time object detection for forward collision mitigation. Read this
case study to understand how we were able to optimise a radar system for low latency and exceptional object detection performance in real time
Interesting post, thanks for sharing.