Participate for the HTC R10 Humanitarian robotic competition which was held in parallel to the HTC R10 conference. A solution to the non ferrous material classification in the metal recycling industry was given by using the research.
The World Bank estimates that, the total solid waste generation per day in Sri Lanka is 23,028 tons. Solid waste in two forms.
Recyclable
Non - Recyclable
Garbage recycling is such a good solution not only for the environmental pollution but also to preserve the natural resources. Recyclable category includes paper, metal, glass, plastic, etc. Although almost every kind of metal can be recycled, only 30 % of the metals are recycled due to reasons such as cost and process complexities.
In our solution we have introduced ,
* A robot for the material classification.
* It consists of a BIONIC SENSING ANTENNA.
* The antenna can identify the material type which comes in contact with it.
* More accurate than existing solutions.
* Fully autonomous solution.
* Operation is relatively fast.
* Material prediction was done by using an Artificial Neural Network (ANN)
#Merits
With remote control facility, METABOT can be used in hazardous environments
High accurate classification results
#Demerits
Accessibility is limited by the mobile robot capability
Full length of the antenna can not be used.
The target is to classify 5 materials (Aluminum, PVC, Wood, Brass and Steel) separately. As the initial step it has to initialize the dataset only for two materials (Aluminum ans Brass). For the input dataset the frequency spectrum (Without dc component) was fed to the neural network. Therefore the number of features of the dataset was 256 frequency components, as an extra feature the contact distance was inserted to the input data. At the beginning dataset was fed without any further changes to the data set. For real time implementation python tensorflow was used with the above dataset. When using python tensorflow some techniques were used to reduce the computations, time consuming.