Revenue Generator

 

  • Technology Licenses- Perpetual or Annual, standalone or customization. e.g.automotive, energy, defense industry.
  • Managed Services in association with partner organizations- on hire, price point based on activities involved. e.g. condition monitoring of infrastructure, warehouse, assets, oil & gas, power, agri-tech, manufacturing.
  • Technology Stake- selling off the technology, IP, software library integration or

 

Competence or Resource Alignment

 

  • Integration of the application library with the customer framework.
  • Testing of the application with the customer data and performing live validation.

 

Value Proposition

 

  • Platform approach allows us to configure various components of solutions to suit
  • specific requirement, thus reduce complexity of development, reduces time and effort
  • to build solutions. This also provides price advantage in go to market.
  • Salient feature like small code footprint, low processing, less memory needs makes it
  • ideal for edge applications.
  • Partial training and integrated learning reduces need of huge dataset(reduces total cost
  • of ownership for customers) for training the algorithms as well as improves accuracy.
  • Multiple input fusion allow customers to use same platform for day, night or twilight
  • vision. Camera based solution makes it affordable in caparison to LiDAR, RADAR,
  • RTK-GPS, UV- Spectroscope or other costly devices. This reduces overall systemic
  • complexity and thus reduces capex and opex related with overall system.

 

Competitive Advantage

 

  • Works on low cost mono camera for FOV object detection, stereo cameras for range SLAM and ultrasonic sensors for proximity MOT. {LIDAR, RADAR}, RTK-GPS and Thermography based solutions are currently expensive and have limitations; Autoliv, SATIR, Sofradir, Bosch, Continental AG, Denso, FLIR, Raytheon, Elbit Systems.
  • Low power requirements, algorithms are optimized for ARM CPU A7 4 core 64 or Intel i5 32 bit 4 GB RAM having installations of Windows, iOS and *ix as compared to many other expensive FPGA or GPU based RNN, CNN, DNN solutions.
  • Algorithm requires limited or partial training, has a high recall rate (error rate ~ 0.458 per 10,000 data points on KMU).
  • Inclusion of Dimensionality in the Vision Platform Stack , Ensemble Algorithms for Object Perception, DARPA MIT LiDAR $10, Tesla Vision.