Page 86 - ESAM-1-4

Engineering Science in
Additive Manufacturing TwinPrint: Dual-arm robotic bioprinting

ballpoint pens were used to determine: (i) Parsing with
respect to two arms, and (ii) setting of global points.
While the parsing process was found to match the
previous two modes, it was noted that there was a position
inaccuracy when each arm would return to the set global
position. Since the global point coordinates are received
as feedback from the arm about its current location
in the 3D space, it was noticed that when the arm was
returned to the initial setpoint, it was slightly off from
the specified point with a deviation <1 mm. This could
be due to mechanical constraints of the arm. Several
models of the Dobot Magician were cross-tested, and
®
similar results were found. However, it is worth noting Figure 6. Estimated errors in flow rate by SmartFlo and TwinPrint
software
that the line thickness of soft matter bioink is generally
>1 mm, as compared to the 0.5 mm line thickness of
a fine-tip ballpoint pen. Thus, it was deduced that the layers approach was used to observe the interaction of
accuracy achieved was suitable for the 3D bioprinting both arms in processing commands sequentially, avoiding
requirements in our scope. collision, and recalling start/stop positions.
To validate layer allocation for each robot, several Several trials were performed to resolve bugs in the code,
combinations of layer splitting were tested for objects which initially caused R2 to repeatedly “forget” its start
of various layer heights. Some combinations required position. Another issue faced was the timeout of the robots
alternate layering, while others required sequential due to extended wait periods in between commands. This
layering (where R1 prints X number of layers and R2 prints was resolved by adding a minimal movement command
Y number of layers, one after the other). The tests were all where R1 would move slightly while in wait position, as R2
found to pass successfully, with both robots able to print completed its layer, and vice versa.
their allocated layers as specified by the user.
For system performance evaluation, the TwinPrint
3.3. Pump accuracy test system was used for 3D bioprinting of an acellular 3D
construct. A seven-layer cuboid of 10 × 10 × 1.4mm was
3
A pump accuracy test was run with three pumps for a range loaded and parsed by the software. Alternating layers were
of 4 flow rates (20, 50, 100, and 200 µL/min) to compare assigned to each robot by splitting them singularly. Pumps
the accuracy of TwinPrint with its default SmartFlo ® were set to 60 µL/min and 20 µL/min for IVZK peptide
software and determine the error. Multiple pumps were and 5× PBS, respectively, based on printing parameters
used to minimize error readings from a single pump. optimized in previous reports. 23-25 R1 was manually moved
Each pump was run for the range of selected flow rates, to a desired start point with a z-height of 0.2 mm from the
first by issuing a command from the SmartFlo software, printbed, and the coordinates were saved using the GUI.
®
and second by TwinPrint. Each test was conducted thrice. The same was repeated for R2 at the same point. The system
Assuming the density of water is approximately 1, the mass was prepared to print. A green dye was injected into S2
of a weighing boat was recorded before and after collecting P6 (5× PBS) to facilitate observation of layers deposited by
the desired volume of water for a period of 1 min. This each arm.
roughly provided the mass of the accumulated volume of
water in the boat. From this, an approximate comparison Given the nature of soft matter ultrashort peptide-based
was deduced of each pump’s performance when receiving bioinks, it was inevitable that accuracy would be reduced
a command from the two software programs. Figure 6 as compared to the pen test. However, printing resolution
shows the results and the estimated error readings. Slight was assessed in terms of a standardized in-house printing
variances were noted between the two systems, but the rubric, which entails observing shape features, continuity
difference was acceptable for the required application. of peptide gel, consistency of layer-buildup, and overall
resolution compared to the desired G-code. Figure 7 shows
3.4. Acellular bioprinting and print resolution test a seven-layer cuboid (10 × 10 × 1.4 mm ) printed with
3
For a thorough demonstration of the TwinPrint system, an both arms depositing alternate layers of peptide-based
experiment was conducted to observe the synchronization bioink. R1 was set to extrude clear peptide ink while R2
of the arms while printing peptide-based 3D constructs. was set to extrude green-stained peptide bioink, allowing
Shape fidelity and cell viability were assessed. An alternating differentiation between the two arms and simulating

Volume 1 Issue 4 (2025) 9 doi: 10.36922/ESAM025410025