Overview

Description

Autobrains' AI Solution is uniquely platform and camera/sensor agnostic, with the critical advantage of low compute/power requirement, which translates to lower cost.

Autobrains provides to people more autonomy. More control, more choice, more freedom. Combined with R-Car, AutoBrains is building a safer, smarter and greener future of mobility for everyone.

Autobrains' unique self-learning AI is modelled on the connectivity of the human brain and is created by the best minds in AI and automotive. It thinks like a human, without human intervention, offering truly innovative technology to OEMs and safer journeys to customers.

The Autobrains open platform makes revolutionary technology, efficiency and agility accessible to all and our people-first approach ensures collaborative outcomes.

Unlock the freedom that comes from safety and discover the potential of every journey with Autobrains.

Autobrains & Renesas’ Tech Competitive Edge vs Competition

Apples-to-apples advantages:

  • Limited / No labeled data is required for learning – quick development/ validation cycles
  • Accuracy – dramatic advantage, especially in tough environments and edge-cases – Proven on 1m miles by a 3rd party
  • Low compute (x10 vs competition) – proven on R-Car V3M Renesas SOC

Features

  • No change in neural weights is required for addition of new objects/concepts – consistency in existing functions
  • Easy and online updatability – signatures can be added (also over the air) without changing the neural network
  • Not a black box – knowledge is represented in Signatures and results can be investigated, visualized and explained
  • Efficient V2X communication and intra- vehicle communication using compressed signatures
  • Adaptivity – online optimization for a given environmental conditions by loading relevant signatures set vs one solution fits all of deep learning
  • Higher-level perception tasks, e.g. prediction D2based on contextual and behavioral cues

Applications

Target Devices

Documentation

Type Title Date
Flyer Log in to Download PDF 1.21 MB
1 item

Design & Development

Related Boards & Kits

Boards & Kits

Videos & Training

Autonomous AI Realized with Low Power Consumption on R-Car V3H (Cortica)

Cortica demonstrates at their booth and in-vehicle their unsupervised learning technology using the highly power-efficient CNN IP available in Renesas’ R-Car V3H SoC for premium front cameras, which is also equipped in the R-Car V3M for high-volume NCAP front cameras. Cortica demonstrates excellent detection of objects, whereas the implementation consumes very low power consumption on the scalable Renesas R-Car V3M and R-Car V3H SoCs.

Cartex 4 Portfolio

Cartex 4M is based on a Renesas V3M architecture with at least a 100° FOV. It is composed of three kernels: NCAP. NCAP+, and narrow functions. The three kernels utilize most of the Renesas V3M resources with the following distribution: NCAP kernel 35%, NCAP+ 25%, and the remaining 40% can be utilized by Cartica’s narrow function. The NCAP kernel requirements are set to allow full scores in NCAP VRU, C2C, and SA test cases - required for the overall 5 stars rating.

Cartex 4H is based on a Renesas V3H architecture, with at least a 100° FOV. It is composed of four kernels: Base, safety, performance, and adventure. The four kernels utilize most of the Renesas V3H resources. The base kernel requirements are set to allow full score in NCAP VRU, C2C, and SA test cases which is needed for the overall 5 stars rating. In addition to other base functionality such as: ACC/TJP, LCA, HLB, ADB, TFL, and free space.

Cartex 4F is based on a Renesas V3H architecture with at least a 100° FOV. It is composed of two kernels: Base, and fusion. The two kernels utilize most of the Renesas V3H resources The base kernel requirements are set to allow full score in NCAP VRU, C2C, and SA test cases which is needed for the overall 5 stars rating. In addition to other base functionality such as: ACC, LCA, TJP, and Traffic light assist.