Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - BINPICKING 3D (Generalized bin picking system for automatic handling of unsorted parts in industrial applications)

Teaser

The ability of humans to manipulate a wide range of objects with great dexterity and precision enables them to operate productively in the world. In a context of ageing population, Europe needs to increase productivity in manufacturing. A strong and competitive industry...

Summary

The ability of humans to manipulate a wide range of objects with great dexterity and precision enables them to operate productively in the world.
In a context of ageing population, Europe needs to increase productivity in manufacturing. A strong and competitive industry creates spillover effects through the entire value chain and value network.
European industry is facing competition from low cost countries. In order to compete successfully, the European companies must allocate its highly skilled workers in high value added tasks and away from repetitive menial tasks ones. The introduction of new smart automation technologies is key to achieve this goal.
In most industries, there are lots of tasks related to handling working parts. One category of these tasks are bin picking, i. e. taking workpieces out from a container or bin. The lowest cost way to move low cost parts between manufacturing activities stores these parts without order in the container. Until now there haven’t been robust technologies to tackle this challenge and random bin-picking solutions have left 5-10% of parts in the bin, a totally unacceptable rate1.
Within this context, the project is based on developing a generic and universal bin picking solution enabling industrial robots to see, find and grip components randomly placed in bins. In contrast to traditional approaches, our innovative approach boosts the perception performance to a manufacturing level by locating the sensor in the robot’s gripper. The technical innovative aspects of this solution comprise: Use of intelligent and adaptive exploration trajectories; Use of hundreds of images of the scene; Estimation of robot encoders permitting a robot independent solution.
Random bin-picking is the ability of a robot to grasp parts from a container without auxiliary alignment devices. It consists of three activities:
1. Isolating objects from the background image.
2. Calculating the objects position relative to the sensor and robot.
3. Generating a path trajectory for the robot to move and grasp the object.
Our goal is to bring our universal bin picking solution from its present TRL6 demonstrator to TRL9 competitive in manufacturing ready to market product. SME instrument phase 1 aims to determine the technological, practical and economic viability of our business idea, together with the resources needed to implement it.

Work performed

InPicker (commercial name of the software product) project has demonstrated the solution prototype in a relevant environment (TRL6). During the business plan elaboration a specific proposal to delineate the approach, activities, resources, and forecasted results for customers, shareholders and society derived from widespread adoption of the solution in manufacturing facilities
From the information obtained during the previous activities, the marketing and commercialisation strategy was defined in order to reach those markets defined as targets with our new product, especially European markets.
A detailed operational plan was defined once the business objectives were set. In this stage we have analysed both the new resources of operational structure needed, and the investment requirements to suit the production capacity.
Financial plan was developed to analyse the financial feasibility of the present project. We have exhaustively calculated the required investment for both the initial development and the production capacity adaptation of the company to the commercial objectives of the new product throughout time.
A market analysis to reinforce the knowledge about the new markets where the company envisages its introduction.
In the next step INFAIMON must execute the following activities:
• Product launch to market.
o Perform pilot implementations in customers.
o Refine final product
o Obtain funding to perform the next project phases.
o Protect IP.
o Perform communication campaigns to gain solution awareness and positioning.
o Obtain distributors.
• Company infrastructure deployment.
o Develop team and organization to support new products and services.
o Deploy infrastructure.

Final results

Trends in manufacturing technologies and initiatives are placing increased importance on the integration of robot technologies in manufacturing facilities (Technology Readiness Levels for Randomized Bin Picking, http://nvlpubs.nist.gov/nistpubs/ir/2012/NIST.IR.7876.pdf ).
InPicker value proposition will be focused on:
• High quality of bin picking performed (as % of parts incorrectly picked).
• Low cost of TCO (low cost of deployment, operation and maintenance, speed of deployment of bin picking, high throughput (high speed bin picking and around the clock production) and reduced labour costs).
• Service support reliability.
• Improved worker\'s safety as they are replaced from highly repetitive tasks as well as those in hazardous conditions.
The results of the technical and economic feasibility study are highly satisfactory. InPicker includes a software system for recognizing workpieces in a container and deciding the better candidates to be picked. The differentiation of InPicker consists in the use of SLAM technology, as well as a different architecture for placing the vision sensors, that allow the generation of hundreds of different images that provide more information from which discriminate the parts and achieve virtually flawless recognition of parts, a problem with other solutions.
The solution proposed represents an important advance over the state of the art but there are limitations to the range of parts that can be recognized effectively by the system. The present version of InPicker could pick about 50% of the possible parts when all the different types of parts are worked into the system. The characteristics of the other 50% of parts (not rigid, texture and light reflection properties, difficult to identify characteristic features,…) do not allow proper recognition with the present state of art.
Around InPicker other products and services can be provided as the InPicker vision head, maintenance and support services, engineering and technical services regarding specific customer requirements as well as customization of recognition of new parts.
The solution should bring €34,4 million in 2025 revenues, €21,9 million in EBIT (a 63,9% margin, typical of software products), the employment of 140 highly skilled people, mainly engineers, as well as a 46,7% in IRR from capital invested.

Website & more info

More info: http://www.inpicker.com.