Improving Workplace Safety with the Parallax PropellerApr 24, 2023
For many, 2020 was quite the decade. While we all tried to survive a deadly virus and out of control wildfires, businesses still need engineers, and startups still have ideas. One of these clients is Occumetric, who contracted us to design and develop their lift test machine.
The project was quite an interesting case study in the use of parallel processing and real time averaging to improve the accuracy of sensor suites, with a bit of biomimicry thrown in.
The concept of Occumetric’s business is to provide machines capable of simulating specific physical tasks for workers, so that their physical capabilities can be safely evaluated. Their first machine was a lifting and lowering test that allowed companies to determine if a person was capable of lifting a required load safely, without overexertion. In the event of a failed lift, the machine would be able to control the load, to avoid any potential risk of injury to the candidate. The machine would need to support lifts from a variety of start positions, and be able to adjust its resistance in order to simulate heavy lifting jobs.
Robots Everywhere was tasked with developing the power and control systems for the machine, receiving task information from an external computer and translating it into control inputs for the lift sled. For this task we chose the Parallax Propeller microcontroller, due to its capability of processing multiple sets of telemetry signals in parallel. Primarily, this was chosen for performance reasons - the position of the sled and the state of the lift must be recorded as precisely as possible - but it also offered improved safety due to the ability to receive emergency signals from the sled, the user, or the control console without any possible latency caused by serial polling. All of this data was then packaged and returned back to the control console.
To avoid the issue of mechanical wear on a serial cable attached to the moving lift sled, we developed a wireless power transmission system using the guide rails and a low voltage current to power a battery operated controller. Data was then sent from the sled to a second Propeller board handling the remainder of the sensors and a high-fidelity wired connection to the control console. Because of this design decision, no wear issues have developed on Occumetric lift test machines over the life of the product to date.
Much like a living organism, to improve the reliability and precision of knowing the exact state of the lift at all times, multiple sensors were used in parallel. By combining a suite of drastically different sensors, translating their outputs into real world values, and then combining them based on performance and interference data, a much more reliable measurement of force and position can be made. For example, ultrasonic rangefinders are more precise than lasers at close distances, but laser rangefinders have a longer operating range. By using an averaging algorithm based on the expected performance curve of each sensor, as well as measured interference, we were able to greatly improve the precision of the instrument.
The greatest challenge of developing this system was developing the data model used to construct the hybrid sensor system. Each sensor in the system needed to be individually tested to know its optimal operating conditions, as well as the type of data returned when the sensor was experiencing interference. Any safety sensors used specifically to detect interference would also have to be included. Once this was determined, these optimums were used along with a moving average to get much more exact position and load data than a single sensor alone would be able to achieve.
After initial deployment, the machines have been reported to be working very well without issue, and we are proud to say Occumetric Services is another satisfied customer for Robots Everywhere.
If you have an idea for a project, and need engineering help, Contact Us. No project is too big or too small.
What can we invent together?