Table 4.
Examples of AI applications for in situ monitoring and in-line correction in 3D bioprinting.
| Process category | Sensor | Input | Output | AI Model | Type of AI Model | Ref |
|---|---|---|---|---|---|---|
| EBB | Camera | Video frame | Printing status | CNN | CQA prediction model | [234] |
| Camera | Video frame | Extrusion status | LSTM autoencoder | CQA prediction model | [302] | |
| Camera | Image | Printing anomaly | CNN | CQA prediction model | [269] | |
| Camera | Video frame | Velocity of the printing head, offset from the baseline printing path | Reinforcement learning | Control Strategy | [271] | |
| Infrared thermocouples | Three features extracted from the raw sensor signals | Printing status strand width, strand height, strand fusion severity | KNN, SVM, RF, ANN | CQA prediction model | [303] | |
| Material temperature, extrusion pressure, print speed, the location in the strand | Regime classification, width prediction, height prediction | |||||
| Stereo cameras | Surface shape data | Shape basis vectors | PCA | CPP prediction model | [292] | |
| Camera, pressure sensor | Time-varying 2D printing head position (X, Y), SMAMs pressure (p1, p2, p3, p4) | The displacement of syringe plungers (l1, l2, l3, l4) | ANN | Control Strategy | [297] | |
| DBB | Camera | Video frame | Droplet evolution in the printing process | Deep recurrent neural network (DRNN) | Process prediction model | [273] |
| Camera | Video frame | Droplet evolution in the printing process | Network of tensor time series (TTS) | Process prediction model | [274] | |
| Camera | Video frame | Jetting status | MovileNetV2 | CQA prediction model | [270] | |
| Camera | Droplet velocity at two different points | Cell count | LR, SVR, decision tree regressor (DTR), RFR, extra tree regression (ETR) | CQA prediction model | [304] | |
| Camera | Droplet size, aspect ratio, droplet velocity, satellite droplet | Droplet mode | Backpropagation neural network (BPNN) | CQA prediction model | [305] | |
| FPGA module for self-sensing signals acquisition | Two features extracted from the raw sensor signals | Nozzle jetting status | SVM, ANN, Gaussian naïve Bayes model | CQA prediction model | [306] | |
| LBB | Camera | Video frame | Printing status | CNN-LSTM | CQA prediction model | [268] |
| – | Digital Mask | Digital Mask | Reinforcement learning | Control Strategy | [272] |