Energy-Efficient Design of an Ostrich-Inspired Bipedal Running Robot for Inspection Tasks

Background & Current Research

During manual inspection of enclosed or hazardous spaces, such as chemical tanks or mine tunnels, workers are often exposed to significant risks, including toxic gas inhalation, asphyxiation, and electrical hazards. These conditions make industrial inspection tasks extremely hazardous, leading to 100 to 200 fatalities every year. Beyond safety risks, manual inspections are often inefficient due to safety protocols and specialized training required beforehand, which can lead not only to delayed detection of infrastructure damage but also to significant labour and time requirements.

Current inspection robots, often tracked or wheeled, struggle on uneven or uncontrolled terrain, such as mud or debris, limiting inspection flexibility and increasing the likelihood of failure. On the other hand, mainstream legged robots, which rely on rigid actuation and high-bandwidth feedback control, are often energy-intensive and have short battery life, requiring multiple charging sessions. In addition, they also cause significant environmental disturbance during operation due to stiff contact dynamics.

Inspiration

I got my idea from observing ostriches and pondering what enables their speed and agility, particularly how their leg locomotion depends heavily on passive compliance in their tendons.

Objective

The objective of this project is to design a bipedal running robot that meets the three requirements:

Energy efficiency through reduced actuation and sensing

Lightweight design

Minimal environmental disturbance for inspection of ecologically sensitive areas

The third and final design of the robot (OBR-III) features one dual-axis DC motor for leg actuation and one servo for the head, which swings to maintain balance. All mechanical parts are 3D-printed in PETG.

Leg Design

Two four-bar mechanisms were used on each leg to drive the leg movement, one replicating the oscillation of ostrich's femur and the other replicating the oscillation of the ostrich's tibiotarsus, as shown in Figure 6. Figure 5 demonstrates how two springs (one in front and one behind the leg) coordinate the leg moment. During stance phase, the springs lock the joint in place, allowing it to support the body weight. During swing phase, the springs retract the leg while flexing the toes backward.

Sensor Module

The robot is also equipped with a camera module and a sensor module, which can monitor ambient temperature, humidity, and the concentrations of oxygen, carbon monoxide, hydrogen sulfide, and methane in real time. The module supports wireless connectivity via STP or AP mode.

Design Iteration

Model 1 (Single-motor Actuation):

The first generation adopted a single-motor actuation scheme and uses 0.6mm thin rope to mimic ostrich tendons. However, this model cannot operate autonomously on the support structure without external assistance, as thin ropes are prone to loosening or entanglement during locomotion.

Model 2 (Dual-motor Actuation):

The second generation utilized dual-motor actuation and PID control architecture was implemented for synchronization. However, insufficient single-leg torque makes it easy for the legs to synchronize rather than alternate, and encoder noise and voltage drop reduce PID control accuracy.

Model 3 (OBR-III):

The third generation abandoned the use of thin ropes and replaced them with tension springs and cable ties, which are not only more durable but also more flexible because the tension can be adjusted through simply finding springs of different stiffnesses.

Experimental Setup

Experiments were conducted on a custom support structure consisting of 8 aluminum profiles and 3D-printed connectors. Red sliders are mounted on the aluminum profiles, allowing the robot to slide horizontally. A vertical metal rod connects the robot body to the experimental setup through a linear bearing. Two experiments were conducted on the robot: speed matching on a micro treadmill, and continuous running on the floor to evaluate the robot's stability and ability to run continuously without lateral deviations or falling over. Through various optimizations during the experiment, OBR-III achieves a stable walking motion at 0.3 m/s with minimal vibration and improved gait balance. In addition, OBR-III also achieves a lightweight design with a total mass of around 1.3kg. Moreover, it achieves minimal environmental disturbance during motion, with an operating noise of 59.4 dB and a ground contact pressure of 24.5 kPa, thereby minimizing surface damage and acoustic disruption during inspection of sensitive environments.

Figure 11 shows the horizontal and vertical displacement of OBR-III as it runs continuously on the floor, while Figure 12 shows its horizontal and vertical velocity. As shown in the graphs, OBR-III starts smoothly and, at maximum speed, runs stably with little vertical fluctuation.

Performance Comparison

OBR-III achieves a cost of transport (CoT) of 0.389 as shown in Equation 1. The CoT of ostriches is estimated to be around 0.2-0.4, which is comparable to the CoT of OBR-III, indicating that OBR-III demonstrates high energy efficiency during locomotion and successfully captures the energy-conserving systems observed in ostriches. Therefore, OBR-III is suitable for long-distance inspection tasks that require energy efficiency and minimal maintenance.

As demonstrated in Table 1, compared to mainstream bipedal robots or inspection robots, OBR-III shows significantly improved energy efficiency, as evidenced by its lower CoT.

As shown in Table 2, OBR-III also achieves improved performance in lightweight design, reduced actuation, and minimal environmental disturbance during operation compared to mainstream bipedal and inspection robots.

Conclusion

In conclusion, this work presents a novel approach to environmental monitoring and inspection of confined or hazardous spaces that challenges the conventional legged-robots design philosophy of relying on extensive sensing, feedback control, and joint-level actuation, and offers a groundbreaking solution for energy conservation in robots. Drawing inspiration from the ostrich's leg mechanism, the design uses tension spring systems and passive compliance to significantly improve the energy efficiency of inspection robots, enabling them to operate for extended periods without frequent charging.

These mechanisms allow the robot to absorb irregular contact, improve terrain adaptability, and reduce control complexity. As a result, this method further improves performance in lightweight design and reduces environmental interference. Ultimately, the results demonstrate that passive compliance can significantly reduce control complexity while enhancing energy efficiency.

On a practical level, OBR-III improves worker safety by providing a mobile robotic platform that reduces the need for human workers to enter confined and hazardous spaces such as tunnels and chemical tanks, eventually lowering the rate of workplace accidents. More broadly, this work contributes to the development of sustainable and accessible robotic technologies.

Future Improvements

Improve the robot’s ability to recover after a fall for better compatibility with complex terrain, potentially using the head as a supporting structure

Incorporating high-energy-density batteries to extend operation time

Integrating depth cameras or reinforcement learning methods for autonomous navigation

Adding further environment-specific sensors, such as pH or PM2.5

Next Steps

Next, I aim to validate this system through real-world testing in environments such as tunnels and industrial pipelines. I also plan to partner with industry partners to bring this solution to market, spreading its impact and reducing energy consumption across the industry.

I would like to thank my parents for their unwavering support throughout this process. I would also like to thank Ms. Wharton and Mr. Chang for their help as I navigate this project.

Thank you to GVRSF for giving me this incredible opportunity of sharing my work with a broader audience.

[1] D. Hoeller, N. Rudin, D. Sako, and M. Hutter, “ANYmal parkour: Learning agile navigation for quadrupedal robots,” Sci. Robot., vol. 9, no. 88, p. eadi7566, Mar. 2024, doi: 10.1126/scirobotics.adi7566.

[2] M. Hutter et al., “ANYmal - toward legged robots for harsh environments,” Advanced Robotics, vol. 31, no. 17, pp. 918–931, Sep. 2017, doi: 10.1080/01691864.2017.1378591.

[3] G. Bledt, M. J. Powell, B. Katz, J. Di Carlo, P. M. Wensing, and S. Kim, “Mit cheetah 3: design and control of a robust, dynamic quadruped robot,” in 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Madrid: IEEE, Oct. 2018, pp. 2245–2252. doi: 10.1109/IROS.2018.8593885.

[4] “Atlas,” Boston Dynamics. Accessed: Oct. 04, 2025. [Online]. Available: https://bostondynamics.com/atlas/

[5] S. Kuindersma et al., “Optimization-based locomotion planning, estimation, and control design for the atlas humanoid robot,” Auton Robot, vol. 40, no. 3, pp. 429–455, Mar. 2016, doi: 10.1007/s10514-015-9479-3.

[6] C. L. Vaughan, “Theories of bipedal walking: an odyssey,” Journal of Biomechanics, vol. 36, no. 4, pp. 513–523, Apr. 2003, doi: 10.1016/S0021-9290(02)00419-0.

[7] M. A. Daley, “Understanding the agility of running birds: Sensorimotor and mechanical factors in avian bipedal locomotion,” Integrative and Comparative Biology, Jun. 2018, doi: 10.1093/icb/icy058.

[8] “Agility robotics.” Accessed: Oct. 04, 2025. [Online]. Available: https://www.agilityrobotics.com/

[9] B. He, S. Wang, and Y. Liu, “Underactuated robotics: A review,” International Journal of Advanced Robotic Systems, vol. 16, no. 4, p. 172988141986216, Jul. 2019, doi: 10.1177/1729881419862164.

[10] N. U. Schaller, B. Herkner, R. Villa, and P. Aerts, “The intertarsal joint of the ostrich (struthio camelus): Anatomical examination and function of passive structures in locomotion,” Journal of Anatomy, vol. 214, no. 6, pp. 830–847, Jun. 2009, doi: 10.1111/j.1469-7580.2009.01083.x.

[11] J. R. Hutchinson, J. W. Rankin, J. Rubenson, K. H. Rosenbluth, R. A. Siston, and S. L. Delp, “Musculoskeletal modelling of an ostrich (Struthio camelus) pelvic limb: influence of limb orientation on muscular capacity during locomotion,” PeerJ, vol. 3, p. e1001, Jun. 2015, doi: 10.7717/peerj.1001.

[12] R. Zhang et al., “Phalangeal joints kinematics during ostrich (Struthio camelus) locomotion,” PeerJ, vol. 5, p. e2857, Jan. 2017, doi: 10.7717/peerj.2857.

[13] D. Gangl, G. E. Weissengruber, M. Egerbacher, and G. Forstenpointner, “Anatomical description of the muscles of the pelvic limb in the ostrich(struthio camelus),” Anat Histol Embryol, vol. 33, no. 2, pp. 100–114, Apr. 2004, doi: 10.1111/j.1439-0264.2003.00522.x.

[14] J. Rubenson, D. G. Lloyd, T. F. Besier, D. B. Heliams, and P. A. Fournier, “Running in ostriches (struthio camelus): three-dimensional joint axes alignment and joint kinematics,” Journal of Experimental Biology, vol. 210, no. 14, pp. 2548–2562, Jul. 2007, doi: 10.1242/jeb.02792.

[15] J. Yu, J. Hooks, S. Ghassemi, and D. Hong, “Exploration of turning strategies for an unconventional non-anthropomorphic bipedal robot,” in Volume 5B: 41st Mechanisms and Robotics Conference, Cleveland, Ohio, USA: American Society of Mechanical Engineers, Aug. 2017, p. V05BT08A021. doi: 10.1115/DETC2017-68051.

[16] J. C. Dean and A. D. Kuo, “Elastic coupling of limb joints enables faster bipedal walking,” J R Soc Interface, vol. 6, no. 35, pp. 561–573, Jun. 2009, doi: 10.1098/rsif.2008.0415.

[17] J. Reher and A. D. Ames, “Dynamic walking: toward agile and efficient bipedal robots,” Annu. Rev. Control Robot. Auton. Syst., vol. 4, no. 1, pp. 535–572, May 2021, doi: 10.1146/annurev-control-071020-045021.

[18] “Passive bipedal running,” Proc. R. Soc. Lond. B, vol. 240, no. 1297, pp. 107–134, May 1990, doi: 10.1098/rspb.1990.0030.

[19] J. W. Grizzle, J. Hurst, B. Morris, H.-W. Park, and K. Sreenath, “MABEL, a new robotic bipedal walker and runner,” in 2009 American Control Conference, Jun. 2009, pp. 2030–2036. doi: 10.1109/ACC.2009.5160550.

[20] A. Badri-Spröwitz, A. Aghamaleki Sarvestani, M. Sitti, and M. A. Daley, “BirdBot achieves energy-efficient gait with minimal control using avian-inspired leg clutching,” Sci. Robot., vol. 7, no. 64, p. eabg4055, Mar. 2022, doi: 10.1126/scirobotics.abg4055.

[21] S.-S. Ha, J.-H. Yu, Y.-J. Han, and H.-S. Hahn, “Natural gait generation of biped robot based on analysis of human’s gait,” in 2008 International Conference on Smart Manufacturing Application, Apr. 2008, pp. 30–34. doi: 10.1109/ICSMA.2008.4505607.

[22] J. W. Hurst and A. A. Rizzi, “Series compliance for an efficient running gait,” IEEE Robotics & Automation Magazine, vol. 15, no. 3, pp. 42–51, Sep. 2008, doi: 10.1109/MRA.2008.927693.

[23] “Do robots outperform humans in human-centered domains?” Accessed: Apr. 29, 2026. [Online]. Available: https://pmc.ncbi.nlm.nih.gov/articles/PMC10661952/

[24] “Confined space safety guide: osha compliance & training,” PK Safety Supply. Accessed: Apr. 29, 2026. [Online]. Available: https://pksafety.com/pages/mastering-confined-safety-guide

[25] DezinsAdmin, “Confined Space Fatalities...an updated look at the numbers,” Roco Rescue. Accessed: Apr. 29, 2026. [Online]. Available: https://rocorescue.com/rescue-talk/confined-space-fatalities-an-updated-look-at-the-numbers/

[26] B. C. Koester and 2018 Aug 01, “We must change the statistics of confined space injuries and fatalities -,” Occupational Health & Safety. Accessed: Apr. 29, 2026. [Online]. Available: https://ohsonline.com/articles/2018/08/01/we-must-change-the-statistics-of-confined-space-injuries-and-fatalities.aspx

[27] “10 key statistics of confined spaces revealing major risks.” Accessed: Apr. 29, 2026. [Online]. Available: https://www.canadasafetytraining.com/Safety_Blog/statistics-of-confined-spaces.aspx

[28] C.-C. Chiu, Y.-M. Chang, and T.-J. Wan, “Characteristic analysis of occupational confined space accidents in taiwan and its prevention strategy,” Int J Environ Res Public Health, vol. 17, no. 5, p. 1752, Mar. 2020, doi: 10.3390/ijerph17051752.

[29] A. Arbuckle, “Confined space awareness - surehire.” Accessed: Apr. 29, 2026. [Online]. Available: https://surehire.com/confined-space-awareness/

[30] S. Simmie, “Adopting inspection robots: Perceived barriers and solutions,” InDro Robotics. Accessed: Apr. 29, 2026. [Online]. Available: https://indrorobotics.ca/adopting-inspection-robots-perceived-barriers-and-solutions/

[31] “Snake,” Sfynx Industry. Accessed: Apr. 29, 2026. [Online]. Available: https://sfynx-industry.com/en/snake-confined-space-inspection-robot/

[32] J. Reher, W.-L. Ma, and A. D. Ames, “Dynamic Walking with Compliance on a Cassie Bipedal Robot,” Apr. 25, 2019, arXiv: arXiv:1904.11104. doi: 10.48550/arXiv.1904.11104.

[33] “Confined Space Hazards – four main causes of serious accidents,” Gasmet.de. Accessed: Apr. 29, 2026. [Online]. Available: https://www.gasmet.com/de/blog/confined-space-hazards-four-main-causes-of-serious-accidents/