The bacteria-robot interface is realized as a highly modular feedback system.
A mobile robot’s sensor values alter the environment of E. coli strains that were designed to respond to these changes with increased expression of fluorescent proteins. In turn, this fluorescence is measured and its quantities are translated into robot behaviour.
"ibpA" - Heat Shock Promoter
The heat shock promoter ibpA is controlled by the transcription factor σ 32. In principle, the exposure to high temperatures leads to an increase of σ 32, which subsequently enables heat shock promoters to be recognized by the RNA polymerase. The promoter exhibits a high induction rate and high levels of expression. In our experiment, the ibpA promoter controls the expression of GFP.
"asr" - Acid Inducible Promoter
Promoter activity is controlled by the RstAB System detecting the pH and the PhoRB System activated when inorganic phosphate is rare. Thus, expression only works in low phosphate media (LPM). When grown in LPM and activated by a switch of pH to 5.5 the promoter becomes active and mCardinal is expressed. To enhance expression an extra ribosome binding site (RBS) was inserted between the promoter and mCardinal.
"alx" - Alkaline-induced Roboswitch
Regulation of translation is managed by a pH sensitive riboswitch. The riboswitch itself, a mRNA part 5’ of the RNA coding for our green fluorescence protein mNeonGreen is regulated by a constitutive promoter. Hence regulation of mNeonGreen translation is managed by the riboswitch and subjected to pH. When pH is neutral, the structure of the riboswitch prevents the ribosome from binding to the RBS. When pH rises, the structure of the mRNA changes and allows the ribosome to bind the RBS and therefore translation can start.
Our system consists of several modules, we differentiate them in 3 layers seen in figure 4. The layer on top is the INPUT layer. It is a steady source of medium or variable liquid, which can be changed and is a key component for its variability. The ANALYSE and MAINTAIN layer consists of two elements: the reactor to maintain our culture and two separate measuring units, one of which is the OD 600 measure unit and the other one can be changed as a variable component (see Interaction Modules (IMs) and Fluorescence Measurement Chamber (FMC)). Both the variable liquid and the variable measure unit give us the freedom to change our system for different projects. In the OUTPUT layer we collect the waste from our measure units. Further to obtain a steady OD 600 value the OD 600 measure unit also transports medium containing cell mass to the waste to regulate the OD 600 value, if it is too high.
Interaction Modules (IMs)
An IM receives bacterial suspension and exposes it to a stimulus, whose quantity is determined by the robot’s sensor readings. We have designed and built two IMs, one for heating up the suspension and a second one to set its pH (both ways). The “Variable Measuring unit” shown in Fig. 4 consists of at least one IM and an Fluorescence Measurement Chamber (FMC). The modularity of our design allows us to replace one IM for another (or even employ both) while keeping the rest of the setup intact.
In this module, tubes carrying the bacterial suspension are coiled between a Peltier-element and a styrofoam block allowing for quick heating of the suspension.
This module consists of two lab bottles, containing acid and base solution. Two peristaltic pumps are controlled according to the sensor values of the robot, to specifically set the pH value of the reactor medium.
Fluorescence Measurement Chamber (FMC)
A 3D printed case housing a cuvette (with in- and outflow tubes), LEDs to excite the fluorescent proteins, optical filters and two camera-modules form our FMC. We look into the raw RGB data in a region of interest to detect bacterial fluorescence. The information from both cameras is aggregated into a single output: the next command to the robot.
A Raspberry Pi coordinates the behavior of all the components of the system as well as offering an interface to users. It maintains a steady-state in the bioreactor, pumps bacterial suspension from the reactor into the other modules (the Interaction Modules (IMs) and Fluorescence Measurement Chamber (FMC) ), controls the behavior of those and communicates with the robot by receiving sensor readings and sending commands.
We hooked up a Raspberry Pi to a Thymio II educational robot in order to enable it to communicate with a server on the internet. The robot moves around in a maze and streams its various sensor readings to the server, while in return receiving commands from the server (according to bacterial fluorescence) to change its behavior.
The arena is one of the parts of our project, which is highly variable. With the help of our model we could show some different scenarios, which are feasible in different arenas. Practically, our experiment was carried out in an arena of various wooden boards with cardboard objects, a collection of wooden boards and obstacles that can be used to set up a great variety of environments for the robot, some simple, others more challenging.