IN VITRO INTELLIGENCE
- General protocol for isolating adipose-derived stem cells (ADSCs): In this project, we used stem cells derived from the artist’s adipose tissue. We washed the fat tissue with phosphate-buffered saline (PBS) containing 5% penicillin/streptomycin (p/s). In case of blood in the tissue, we left it in a petri dish for two hours and then changed the medium to separate the red blood cells from the tissue. We then transferred the tissue to a solution of type XI collagenase (+PBS + 2% p/s) and incubated it for 40 minutes at 37°C. In the next step, we cut the tissue into the smallest pieces possible with forceps and a scalpel, and transferred it back to the incubator for 30 minutes. We mixed the tissue every 10 minutes to speed up the degradation. We then transferred the tissue to a centrifuge and mechanically degraded it further with a pipette. We added an equal volume of stromal medium (DMEM/F12, 10% FBS, 1% p/s) to the centrifuge to deactivate – General protocol for isolating adipose-derived stem cells (ADSCs): In this project, we used stem cells derived from the artist’s adipose tissue. We washed the fat tissue with phosphate-buffered saline (PBS) containing 5% penicillin/streptomycin (p/s). In case of blood in the tissue, we left it in a petri dish for two hours and then changed the medium to separate the red blood cells from the tissue. We then transferred the tissue to a solution of type XI collagenase (+PBS + 2% p/s) and incubated it for 40 minutes at 37°C. In the next step, we cut the tissue into the smallest pieces possible with forceps and a scalpel, and transferred it back to the incubator for 30 minutes. We mixed the tissue every 10 minutes to speed up the degradation. We then transferred the tissue to a centrifuge and mechanically degraded it further with a pipette. We added an equal volume of stromal medium (DMEM/F12, 10% FBS, 1% p/s) to the centrifuge to deactivate the collagenase. We centrifuged the contents at 1000 revolutions per minute (RPM) for 10 minutes, then vigorously shook the centrifuge to mix the contents and centrifuged again at 2000 RPM for 5 minutes. This separated the floating fat tissue from the stromal vascular fraction (SVF). We first removed the fat on top with a pipette, then the remaining liquid. We resuspended the pellet in stromal medium and filtered the cell suspension through a 70 μm cell strainer. We were careful to remove any cell clumps. If they appeared, we shook them (on a vortex) and then injected them three times through needles (from the thickest to the thinnest – needle sizes: 23, 21, 19). We seeded precursor cells from the stromal vascular fraction in T75 flasks at 10,000 cells/cm² in stromal medium and incubated them at 37°C and 5% CO₂. We changed the medium every 2-3 days. After 5 to 7 days, the cells reached 80% growth, which represents confluence. When we reached confluence, we froze most of the cells at -80°C. We differentiated the rest into neurons.
- Neuronal differentiation: ADSCs were grown to 80% confluence in stromal medium. Then we removed the stromal medium, which we stored for later use, and rinsed the cells with PBS and added Accutase, which detaches the cells. We incubated the cells at 37°C for a maximum of 10 minutes. We then added the old stromal medium, transferred them to a centrifuge, and centrifuged them at 1200 RPM for 5 minutes. We removed the supernatant and resuspended the pellet in fresh stromal medium, then seeded the cells on a multi-electrode array (MEA, Multichannel Systems). After 3 to 5 days, we removed the stromal medium, and rinsed the cells with PBS. We then added the neural induction medium (NIM), which contains: DMEM/F12, 10 µM forskolin, 200 µM butylated hydroxyanisole in 0.5% ethanol, 5 mM KCl, 2 mM valproic acid, 1 µM hydrocortisone, and 5 µg/mL insulin. The induction medium needs to be changed every 2 to 3 days. Induction of neurons takes 3 to 5 days
the collagenase. We centrifuged the contents at 1000 revolutions per minute (RPM) for 10 minutes, then vigorously shook the centrifuge to mix the contents and centrifuged again at 2000 RPM for 5 minutes. This separated the floating fat tissue from the stromal vascular fraction (SVF). We first removed the fat on top with a pipette, then the remaining liquid. We resuspended the pellet in stromal medium and filtered the cell suspension through a 70 μm cell strainer. We were careful to remove any cell clumps. If they appeared, we shook them (on a vortex) and then injected them three times through needles (from the thickest to the thinnest – needle sizes: 23, 21, 19). We seeded precursor cells from the stromal vascular fraction in T75 flasks at 10,000 cells/cm² in stromal medium and incubated them at 37°C and 5% CO₂. We changed the medium every 2-3 days. After 5 to 7 days, the cells reached 80% growth, which represents confluence. When we reached confluence, we froze most of the cells at -80°C. We differentiated the rest into neurons.
ARTIFICIAL INTELLIGENCE
- When designing robots with a large number of degrees of freedom, choosing the appropriate morphology and kinematics can be a challenge. The Biobot project solves this problem by using the RoboGrammar approach, which operates in two phases by solving these problems in simulation. In the first phase, an evolutionary optimization algorithm seeks a robot configuration that can move in a given environment. This involves creating different robot configurations from a defined grammar and selecting the most promising ones for evaluation in the second phase based on a learned value function.
- The grammar determines which parts of the robot body can be combined. In the second phase, the Model Predictive Control (MPC) algorithm is used to determine the sequence of movements that maximizes the change in the robot’s position. This algorithm is used to select the most promising sequence of movements within a given time frame. This two-phase process is improved by incorporating neural signals obtained from OpenEphys. Although we still use MPC to find the sequence of movements, action potential frequencies regulate the torque in the joints. This means that smaller or larger frequencies correspond to smaller or larger abilities to move a specific joint. This can affect the robot’s ability to move and the morphology of the robot we are looking for in the first phase of the algorithm.
INFRASTRUCTURE SUPPORTING LIFE
- A specially designed incubator allows us to observe cell growth in real-time, directly through a built-in microscope and computer connected to other data acquisition and visualization systems that are processed in the next phase. The incubator was designed and manufactured as part of the S+T+ARTS residency in collaboration with laboratory equipment manufacturer Kambič. The main difference from other incubators is nine Peltier plates that allow a precise temperature range necessary for growing cell cultures. The hardware also supports the management of an electromagnetic valve to provide a 5% atmosphere of CO2.
- The data acquisition system for reading neural signals consists of hardware, software, and Open Ephys protocols, which provide a cost-effective method for obtaining signals from neurons grown on the MEA technology for in vitro multi-matrix technology. Basic protocols and hardware construction plans provide a cost-effective and versatile accurate amplification system for in vitro electrophysiological research with a multi-electrode matrix. This system also verifies connections between the customized PCB board and the INTAN RHD200-EVAL chip and allows for electroporation, acquisition, and transfer of signals to a computer via a data acquisition electronic component in the Open Ephys system.
- The robot is customized and made of SLA-printed parts by the author and assembled on Dynamixel X-Series servomotors produced by ROBOTIS. The XL330-M288-T servomotor is equipped with contactless magnetic encoders that allow 360° rotation at a maximum of 61 RPM. It can withstand a torque of up to 0.52 Nm at 5V.
