Biosensors featured with high selectivity and sensitivity are ubiquitous for prevailing biological analyses, showing great potential in diverse applications for detecting trace-level chemicals and biomolecules in biomedical engineering, environmental protection, and industrial fabrication, etc. ^[1][2][3][1,2,3]. For the last decades, scientists devoted to this research field have developed a number of innovative biosensors according to the diverse characteristics of tested objects ^[4][5][6][4,5,6]. Recently, it has become possible for biosensors to measure analysts in a real-time manner, which is significant for the practical monitoring of rapid changes in biological liquids ^[7][8][9][7,8,9].

In modern societies, people pay more attention to effective and affordable measurements of food safety [10], disease prevention, health status, and chronic treatment in their daily life [11]. However, for a long time, accurate detections of biological substances can only be performed in laboratories to achieve high sensitivity and low detection limits [1], which requires bulky and expensive systems that can only be operated by professional technicians [12]. Consequently, biosensors that can achieve high-quality sensing with cost-effective implementation have attracted more and more attention. Recently, the combination with personal intelligent terminals such as mobile phones and iPads makes biosensors more powerful in data processing and online diagnosis ^{[13][14][15][16]}[13,14,15,16]. To meet the requirements of portable devices for easy-to-go testing, miniaturization is another crucial research focus in biosensor technology [17]. However, the miniaturization of biosensors often suppresses their detection accuracy because of the unexpectedly low efficiency of passive transport in micro volume samples. Currently, the compromise between the detection performance and miniaturization is still an unmet challenge for developing state-of-the-art biosensors.

2. Artificial Microrobots for Biosensing

As innovative artificial machines in micro/nano scale, microrobots can be driven to operate propulsion on demand under the low Reynolds number constraints and carry payloads to move precisely under a navigation strategy for overcoming the Brownian motion ^[18][34]. With the rise and rapid development of nanotechnology, researchers have been striving to shrink the functionalized robots to cellular and molecular levels in order to perform delicate tasks in micro/nanoscale, through precise monomer or cluster control. Since then, microrobots consisting of inorganic oxides or smart materials have come into the spotlight in the field of diagnosis, sensing, microsurgery, targeted drug/cell delivery, thrombus ablation, and wound healing ^[19][20][21][35,36,37]. In 2016, the Nobel Prize in chemistry highly recognized the great potential of molecular motors, promoting the advancement and innovation of miniaturization technology. Since then, microrobots (also called micromotors, microengines, microrockets, etc.) have come into the spotlight as a powerful tool in various areas including drug delivery ^[22][38], nanosurgery, biosensing, and detoxification, which can convert diverse energies into efficient autonomous movement ^[23][24][39,40]. The detection methods, based on fluorescence, surface-enhanced Raman scattering (SERS), locomotion, electrochemical current (EC), and electrochemical impedance spectroscopy (EIS), have now been successfully combined with micro/nanorobots and are ready to be used in portable devices. Among these methods, fluorescence detection is widely accepted with high sensitivity and selectivity and has great advantages in detecting the presence of biomolecules. The enhanced Raman technique shows great potential in the detection of trace-level objects. The electrochemical method has the advantage of high sensitivity and fast response. Beyond these five principles, it should also be noted that there are additional detection methods such as acoustic sensing and thermal sensing for biological detections. However, these types of biosensors are always used for low sensitive and harsh environments where microrobots cannot be implemented to perform accurate manipulations. The applications of different types of microrobots in biosensing involve five detection principles: fluorescence detection, SERS, locomotion, EC, and EIS. The design and utilization of these microrobots offer innovative approaches to the field of biosensing. In general, the apparent geometries of microrobots are 50 nm~10 μm and their moving speed is 500 nm/s~1 mm/s. A variety of microrobots have been introduced with different materials and principles for distinct target detections. Taking fluorescence detection, for example, the fluorescence intensity needs to reach the lowest intensity for detection, which can be enhanced with the controlled motion of Janus micromotors or graphdiyne tubular catalytic microrobots ^[25][41]. The miniaturization of the fluorescence detection device has been achieved, enabling the entire detection process to be completed on site within 5 min. Microrobots made of precious metals, such as gold and silver, are also excellent probes for SERS detection to keep a close contact with analytes and on-demand enrichment, which enhanced the Raman signal approximately three times ^[26][42]. In addition, monitoring the locomotion of chemically powered microrobots is an effective mean for directly detecting chemical concentration changes ^[27][43]. 

3. Biosensors with Microrobots

3.1. Microrobots in Fluorescent Biosensing

Fluorescence detection integrated with nanomaterials is a widespread approach to detect and quantify various chemicals and biomarkers ^[28][62]. The attraction of fluorescence as an analytical tool lies in the simplicity of detection, where the devices require only a closed light environment and a light detector. Variation in fluorescence intensity and color is achieved by changing the binding state of the probe to the analyte. In divergent detection scenes, various materials and detecting methods are applied in corresponding detection conditions and exhibit different levels for the limit of detection. Among these materials, graphene oxide (GO) or reduced graphene oxide (rGO) are more attractive because they can absorb the dye-labeled aptamer through π−π stacking interactions. The exceptional surface properties of graphene have allowed the attachment of different receptors for toxin detection ^[29][63]. Consequently, these receptors enable the capture of nerve agents and heavy metals ^[30][64], and then the detection can be accomplished by their fluorescence-quenching ability ^[31][32][33][65,66,67]. Additionally, one of the most widely applied fluorescence is based on the aptamer, which is taken as an excellent example of functional molecules selected in vitro ^{[34][35][36][37][38][39][40]}[68,69,70,71,72,73,74]. More importantly, aptamers have high specificity for certain targets, ranging from small molecules to large proteins and even cells, which offers remarkable flexibility and convenience for designing biosensors with high sensitivity and selectivity ^{[41][42][43][44]}[75,76,77,78]. Compared with antibodies or enzymes, aptamers have higher chemical stability and convenience in the structure design ^{[45][46][47][48][49]}[79,80,81,82,83]. With the development of smartphones, fluorescence detection has been freed from the bondage of laboratory microscopes [10] and has become increasingly convenient for the end-users ^[34][68]. Integrated with microrobots for real-time fluorescence sensing of (bio)markers, Yuan et al. described the design of a portable device composed of a smartphone coupled with a high-resolution optical lens, custom-made emission filters, and a compartment for the insertion of low-cost commercial lasers to tailor the excitation wavelength ^[25][41]. Magnetic Janus microrobots modified with fluorescent ZnS@Cd_xSe_1−x quantum dots and graphdiyne tubular catalytic microrobots modified with rhodamine-labeled affinity peptide were, respectively, used for the OFF-ON detection of mercury and cholera toxin B.

3.2. Microrobots in Surface-Enhanced Raman Scattering Biosensing

Raman scattering refers to an inelastic light scattering process that provides a vibrational spectrum representing chemical structure information ^{[50][51][52][53]}[85,86,87,88]. However, the Raman scattering is a weak process, and generally the light intensity is only approximately 10⁻¹⁰ of the incident light intensity ^[54][55][56][89,90,91]. Surface-enhanced Raman scattering (SERS) is based on the enhancement effects of the rough surface of noble metals, which are 10⁴–10⁷ times stronger than traditional Raman scattering signals ^{[57][58][59][60]}[92,93,94,95]. SERS is considered to be promising ^[61][62][96,97] as an unlabeled and rapid biosensing technology with high specificity and sensitivity through the SERS-provided vibrational spectrum information ^{[63][64][65][66]}[98,99,100,101]. To improve the SERS performance, microrobots were attempted to be integrated in pioneer studies. The enhanced caption of SERS signal could be acquired by inducing adequate SERS probes in the detection area and maintaining close contact between the probes and analytes ^[50][85]. Wang et al. presented an active SERS probe of a light-powered micro/nanomotor (MNM) which has the matchlike AgNW@SiO₂ core-shell structure ^[26][42]. The maximum speed for this type of micromotor is approximately 9 μm/s with ~30 μm length and the AgCl tail. According to micromotor enrichment remotely controlled by external light, both 10⁻⁴ M crystal violet and MCF-7 breast cancer cells were successfully detected with three-times-enhanced Raman signal. Using such a light-induced enrichment of the nanomotors, the Raman signals can be enhanced 6.2 times in a localized detection area in microscale as a supplement to the conventional Raman signal enhancement by SERS.  Meanwhile, portable SERS readers for a lateral flow immunoassay (LFA) were proposed to simplify the essential complicated operations in the laboratory. Li et al. proposed a LFA strip based on SERS nanotags for the simultaneous and quantitative detection of dual infection biomarkers, serum amyloid A (SAA) and C-reactive protein (CRP), respectively ^[67][103]. Such a biosensing system achieved LODs as low as 0.1 and 0.05 ng/mL, respectively, for SAA and CRP. Tran et al. presented a Raman/SERS-LFA reader that uses a custom-made fiber optic probe for rapid, quantitative, and ultrasensitive POCT ^[68][104].

3.3. Microrobots in Locomotion-Based Biosensing

The reliance of the microrobot movement on external power sources including both physical fields and chemical fuels makes it possible to design a special category of biosensors based on the relationship between the microrobot moving speed and the change in their surrounding environments ^[69][70][71][105,106,107]. Based on the working principle of microrobot motion-based detection, the concentration of the detected substance will affect the microrobot’s speed, acceleration, or deceleration ^[72][73][74][108,109,110]. Once the linear or non-linear relationship was built, the quantification detection of target reagents could be easily conducted by analyzing the motion of microrobots in real time. Moreno-Guzman et al. reported a one-millimeter-sized tubular micromotor for mobile biosensing of H₂O₂ in environmental and relevant clinical samples ^[75][111]. Sodium dodecyl sulfate (SDS) surfactant and horseradish peroxidase were released from the rear of the microrobot which was propelled by the Marangoni effect. In this case, the motion of a single millimeter-sized tubular micromotor for 120 s was measured to quantify the concentration of H₂O₂ in different samples. Similarly, Orozco et al. presented a novel microrobot-based strategy for water-quality testing based on changes in the propulsion behavior of artificial biocatalytic microswimmers influenced by aquatic pollutants ^[76][112]. The presence of 100 μM Hg leads to a rapidly diminished propulsion efficiency with speed diminutions of 90–95% for enzyme-decorated PEDOT/Au millimeter-sized tubular micromotors within 6 min.  Moreover, the optical equipment is essential for observing microrobot movement. Portable devices with a combination of a smartphone and an optical lens show great superiority compared with the traditional lab-used bulky microscopes. 

3.4. Microrobots in Electrochemical Current-Based Biosensing

Apart from optical methods that rely on the microscope- or smartphone-integrated magnify lenses, electro-signal-based detection offers another option to build portable biosensors independent of the delicate optical system, and thus is highly suitable for incorporating with existing MEMS devices ^{[77][78][79][80]}[115,116,117,118]. The electrochemical detection reads the current signal generated by the redox reaction of the analyte on the electrode to obtain the concentration of the analyte ^[81][82][51,119]. As one of the most important methods in electroanalytical chemistry, cyclic voltammetry (CV) is often applied in electrochemical current detections ^{[83][84][85][86][87][88][89][90][91][92][93][94][95]}[46,120,121,122,123,124,125,126,127,128,129,130,131]. The addition of microrobots in electrochemical detection accelerates the ambient liquid flow, so as to improve the detection sensitivity and LOD. Rojas et al. presented a novel Janus microrobot-based strategy for the direct determination of diphenyl phthalate (DPP) in food and biological samples. Mg/Au Janus microrobots (average diameter, 20 μm) degraded DPP to phenol, which is directly measured by difference pulse voltammetry on disposable screen-printed electrodes ^[83][46]. Similar to the optical counterparts, electrochemical detection based on smartphones has also been reported. Smartphones serving as data processors and displayers in electrochemical detection have a great potential to combine advanced 5G technologies. Ji et al. designed a smartphone-based CV system for portable detection. The system consisted of screen-printing modified electrodes, a portable electrochemical detector, and a smartphone ^[87][123]. The reduced graphene oxide (rGO) and 3-amino phenylboronic acid (APBA) were modified on the screen-printed electrodes for detection. The LOD for glucose was approximately 0.026 mM with test errors less than 3.8% compared with the commercial electrochemical workstation. 

3.5. Microrobots in Electrochemical Impedance Spectroscopy Biosensing

By means of measuring the intrinsic electrical properties of the target in the electric field, electrochemical impedance spectroscopy (EIS) offers a biocompatible and harmless methodology for detecting various items such as bacteria, biological cells, and tissues, etc., ^{[96][97][98][99]}[132,133,134,135]. Though EIS has an unmatched advantage in cell identification by measuring the impedance signal corresponding to distinct cell lines in a biocompatible way, excluding toxic reagents, the precision of detection is highly dependent on the location of the cell during the detecting procedure or the data calibration process. Introducing microrobots and their motion control systems can convert the existing unmanageable cell passage into a controllable route and significantly improve the detection performance. Wan et.al realized an efficient detection of circulating tumor cells (CTCs) through EIS means utilizing Mg-based microrobots ^[100][61]. Using an aldehyde–amine condensation reaction, the Mg-based microrobots can be modified with Fe₃O₄/P/anti-E nanoparticles to capture CTCs. With the hydrogen (H₂) propulsion by Mg reaction, microrobots maintain 16.5 μm/s motion to increase the chance of anti-E recognition and capture of CTCs. The EIS detection platform based on Mg-based microrobots have a good linear response range and low detection limit for CTCs in untreated blood samples (~5 cells/mL). This method was also demonstrated to be effective for detecting oxidized low-density lipoprotein (Ox-LDL) in whole blood just by replacing anti-E components with the antibody of Ox-LDL for microrobots ^[101][136]. Portable electrical impedance analyzers rely heavily on smartphones for data analysis and display. Zhang et al. developed a smartphone-controlled biosensor system that consisted of a miniaturized biosensor, a hand-held EIS detector, and a smartphone to quantify different kinds of proteins by exchanging electrodes for POCT ^[102][137]. The smartphone provides control commands and receives data signals via Bluetooth, meanwhile acting as a displayer for the measurement results in form of a Nyquist plot.