Typical large operations receive raw data streams from tens of thousands of sensors. One must surmount several challenges to turn these raw streams into high-quality data:
Noise in individual sensors, rendering
data imperfect in unpredictable ways
Huge volume of data being generated, at a high rate
Sensor fusion, or how to consistently and
coherently merge data from many sensors
Hidden variable discovery, i.e. recovering
indirect or inferred information from sensed data
High quality data, once obtained, is used for decisions. Long term decisions fall in the domain of optimization. Decisions to be taken almost instantaneously fall under the purview of “control”. Taking and actuating optimal decisions every instant is challenging, requiring careful engineering and advanced mathematical algorithms.
THE RIGHT MIX OF TECHNIQUES
The field of sensing and control evolves rapidly, making it hard to devise optimal solutions. We help clients find the right mix of techniques from the kaleidoscope of options available. And we invent new ones when nothing suitable exists.
EXAMPLE PROJECTS
Multi-axial Material Testing Machine
No machine exists that can accurately measure parameters of mechanical behavior of non-linear materials. So we decided to build one. Though the machine is novel because of a host of algorithms running inside, the sensing & control problem itself was quite tricky: the machine has 13 actuators, 40 sensors, and a vast distributed sensing & control system.
Read more about the machine and its applications