I was wondering, how driverless cars work?. Which PLCs do they use?. SIEMENS, Controllogix, Telemecanique???.. I was astonished to find none. No more PLCs and ladder logic for automation !!!. After doing some deeper research, i realized that PLCs collect process data as DI and AI… A very old fashioned way… But the brain behind the driverless cars collect process data as information. Yes… they see the road like humans and process it. The deep learning algorithm behind self driving autonomous vehicle’s controllers could NEVER have been conceived with primitive data points DI, DO, AI and AO alone. Looks like traditional PLCs are getting obsolete. AI Powered Controllers are taking over. Python is replacing primitive ladder logic as the preferred language for process control and AI. Watch on…
Self-driving taxis hit the streets in Singapore
Self driving autonomous cars is a big blow to the PLC giants. They have realized that this kind of automation is simply impossible with their PLCs. The ground has shifted.
The heart of traditional manufacturing process has always been Temperature, Pressure and Flow. Vision and sound was missing. It was mostly confined to the laboratory analysis (laser analyzers, spectrometer, diffractometer, quantometer, doppler ultrasonic analysis etc). PID controllers are not designed to interpolate the process history and predict the outcome.
Today with the low cost, high power embedded processors, vision, ultrasound analysers and high end control algorithms like adaptive control, autopilot and ML will become integral part of process automation. High availability AI powered controllers will take over automation. PLCs will be relegated as low end data collecting slaves of AI controller.
All physical properties like temperature, pressure, flow, sound, electrons, bosons etc can be measured noninvasively by analysing the light spectrum reflecting or originating from those objects and particles. Of course, space scientists explore planets, stars, galaxies, universe and big bang by spectral analysis. In short, spectrum is the signature of nature.
Of course it’s not possible with standard PLCs. Remote telemetry and industrial machine automation are a completely different world than autonomous vehicles.I’m not sure what your point is, but it seems like you are trying to compare apples and oranges.
I would like to take automation to the next level. Integrate Machine Learning algorithms which can command and control the PLCs. Like this self learning Quadcopters. Do you think, it’s possible?.
It’s absolutely possible. It’s not, however, trivial.
A good place to start is with a thorough understanding of various ML algorithms, their strengths and weaknesses, and how to implement them. Andrew Ng’s free Coursera class (https://www.coursera.org/learn/machine-learning) is a great place to start.
This topic could not possibly be complete without a mention of my Ethernet/IP Class1 Communications Module, which implements enough of a PLC’s architecture within Ignition to not only scan and control I/O devices (with the premium version of the module), but to run jython replacements for the PLC code (called from Gateway timer scripts).
You are right. Actually i don’t want it to be a trivial thing like PLC programming which is being taken over by low cost manpower at an alarming rate. To survive, we need something which will keep this LCMP away for some more time. That’s the “little secret” behind my love for ML and AI
“Embedded Ignition and PLC on ARM Linux + Full stack Python + ML engine” would be a game changer.
Soft PLC on Windows will be a disaster. Soft PLC manufacturers must switch over to Linux, if they want to sell their product. It’s not easy to convince a decision maker to change his mind, when it comes to rock solid stability and reliability of a proven product. The best opportunity existing today is, ML on top of PLC.