A novel algorithm is proposed for the automatic segmentation of diesel spray images and the calculation of their macroscopic parameters. The algorithm automatically detects each spray present in an image, and therefore it is able to work with diesel injectors with a different number of nozzle holes without any modification. The main characteristic of the algorithm is that it splits each spray into three different regions and then segments each one with an individually calculated binarization threshold. Each threshold level is calculated from the analysis of a representative luminosity profile of each region. This approach makes it robust to irregular light distribution along a single spray and between different sprays of an image. Once the sprays are segmented, the macroscopic parameters of each one are calculated. The algorithm is tested with two sets of diesel spray images taken under normal and irregular illumination setups.

Cable-driven parallel robots are a special type of robot in which an end-effector is attached to a fixed frame by means of several cables. The position and orientation of the end-effector can be controlled by controlling the length of the cables. These robots present a wide range of advantages, and the control algorithms required have greater complexity than those in traditional serial robots. Measuring the cable tension is an important task in this type of robot as many control algorithms rely on this information. There are several well-known approaches to measure cable tension in cable robots, where a trade-off between complexity and accuracy is observed. This work presents a new device based on strain gauges to measure cable tension specially designed to be applied in cable-driven parallel robots. This device can be easily mounted on the cable near the fixed frame, allowing the cable length and orientation to change freely, while the measure is taken before the cable passes through the guiding pulleys for improved accuracy. The results obtained from the device show a strong repeatability and linearity of the measures