Emerging 3D Vision Technologies for Industrial Robots

As more indus­tries come to terms with the need for robots in their oper­a­tions, the num­ber of indus­tri­al robots installed world­wide will expe­ri­ence pos­i­tive growth with­in the next five years. With this devel­op­ing inter­est in automa­tion comes more invest­ments into research and devel­op­ment. The out­come is lean­er robot­ic sys­tems with more advanced com­po­nents. A grow­ing trend for robot­ic work­cells is 3D auto­mat­ic vision. This tech­nol­o­gy allows the robot to iden­ti­fy an object’s posi­tion, size, depth, and col­or. Sec­tors like logis­tics, food pro­cess­ing, life sci­ence, and man­u­fac­tur­ing are find­ing ways to auto­mate their process­es using visu­al components. 

What are the dif­fer­ent approach­es to 3D vision?

Vision tech­nol­o­gy is not a ​“one size fits all” tool. Cer­tain fac­tors such as appli­ca­tion, equip­ment, prod­uct, envi­ron­ment, and bud­get will deter­mine how to inte­grate vision into the process. There is no stan­dard when it comes to set­ting up real-time 3D imag­ing in a robot­ic sys­tem. How­ev­er, there are a few stan­dard tech­niques used by vision-inte­gra­tion experts, each tai­lored to ben­e­fit spe­cif­ic tasks. These tech­niques are stereo vision, time-of-flight (TOF), laser tri­an­gu­la­tion, and struc­tured light. 

Laser Tri­an­gu­la­tion

Objects pass through a beam of light emit­ted by a laser scan­ner. A cam­era posi­tioned at a spe­cif­ic angle records an image of the laser line as the item pass­es through, dis­tort­ing the beam and cre­at­ing a pro­file of the object. 

Struc­tured light

A pro­jec­tor cre­ates a thin band of light to project a pat­tern on an object. Cam­eras from dif­fer­ent angles observe the var­i­ous curved lines from the light to devel­op a 3D image of the object. 

Time of Flight (ToF)

A cam­era uses a high-pow­er laser scan­ner to emit light reflect­ed from the object back to the image sen­sor. The dis­tance from the cam­era to the object is cal­cu­lat­ed based on the time delay between trans­mit­ted and received light.

Stereo vision

The robot­ic sys­tem uses two cam­eras to record the same 2D view of an object tak­en from two dif­fer­ent angles. The soft­ware then uses the estab­lished posi­tion of the two cam­eras and com­pares cor­re­spond­ing points in the two flat images to iden­ti­fy vari­a­tions and pro­duce an image.

What appli­ca­tions are using 3D robot­ic vision?

There is a need for the mod­ern indus­tri­al robot to detect objects, rec­og­nize parts, and grip com­po­nents at the right angle. While tra­di­tion­al robots are per­fect for locat­ing parts con­sis­tent­ly, mod­ern robot­ics can coor­di­nate cor­rec­tions to detect where the piece is. Instead of an entire pro­duc­tion line com­ing to a stop because sub­se­quent actions are not indef­i­nite order, the sys­tem quick­ly rec­og­nizes a change and adapts to it. As a result, an array of indus­tri­al appli­ca­tions across indus­tries invest in 3D robot­ic vision. These include the logis­tics, food pro­cess­ing, life sci­ence, man­u­fac­tur­ing, and auto­mo­tive indus­tries. With so many sec­tors automat­ing, the use of vision tech­nol­o­gy is expand­ing into new ter­ri­to­ry. Depal­letiz­ing appli­ca­tions use 3D vision com­po­nents to scan pal­lets filled with var­i­ous types of ship­ping box­es for sort­ing. They use scan­ners to send the image to soft­ware to allow the robot to detect box types based on tex­ture pat­terns and send them to des­ig­nat­ed areas. A food pro­cess­ing plant uses mul­ti­spec­tral vision tech and spe­cial light­ing to inspect the prod­uct and detect spoilage. Appli­ca­tions that have tra­di­tion­al­ly used vision tech­nol­o­gy are upgrad­ing to more inno­v­a­tive equip­ment. An aero­space com­pa­ny replaced tra­di­tion­al inspec­tion tools with 3D scan­ning to inspect tur­bine blades for imper­fec­tions, reduc­ing inspec­tion time from 18 hours to 45 min­utes. Vision tech­nol­o­gy will con­tin­ue to expand, with future trends pre­dict­ed in logis­tics appli­ca­tions, mul­ti­spec­tral machine vision, adap­tion using machine learn­ing with 3D vision, and liq­uid lens­es to allow more pre­cise images from greater distances. 

Cru­cial sub­sys­tems and com­po­nents for vision applications

The most coor­di­nat­ed automa­tion sys­tems have more than a sin­gle auto­mat­ed con­trol sys­tem and com­po­nents inte­grat­ed to make an effi­cient work­cell assem­bly. When it comes to incor­po­rat­ing advanced 3D vision options like object track­ing, prod­uct pro­fil­ing, and bin pick­ing into a process line, the sys­tem should gen­er­ate 3D imagery data. The use of 3D vision in robot­ic sys­tems requires inte­grat­ing var­i­ous com­po­nents to facil­i­tate ade­quate pow­er sup­ply, real-time pro­cess­ing, and safe­ty. Anoth­er crit­i­cal com­po­nent of suc­cess­ful automa­tion is com­mu­ni­ca­tion capa­bil­i­ty. It is good prac­tice in the dig­i­tal age to have con­nec­tiv­i­ty ports to dig­i­tal­ly con­nect a sys­tem to oth­er pieces of equip­ment for data shar­ing. Emerg­ing robot tech­nolo­gies facil­i­tate Wi-Fi con­nec­tiv­i­ty for the same pur­pose. At the design stage, dri­ving a risk assess­ment study is the only way to iden­ti­fy and remove prob­lems from a sys­tem that could risk mal­func­tion. A 3D vision-enabled robot can safe­ly stop equip­ment to pre­vent injury and dam­age to equip­ment. If buy­ers invest in the research and upfront plan­ning, the result will be a flex­i­ble and easy-to-use auto­mat­ed system. 

Con­clu­sion

Mod­ern man­u­fac­tur­ing demands more out of less, with lean­er pro­duc­tion lines need­ing to pro­vide greater out­put. The influ­ence of robot­ic vision will con­tin­ue to expand into dif­fer­ent pro­duc­tion areas and find brand-new ways to improve auto­mat­ed process­es. Expect more 3D visu­al com­po­nents to become com­mon in auto­mat­ic sys­tems in the future.