"Physical Examination Report" for Robot Soles: Performance Test Methods and Evaluation System
- Where the Problem Originates
Let’s start with a typical conversation in the industry.
"How is the wear resistance of this robot sole of yours?""Pretty good. It hasn’t been replaced after three months of operation at the customer’s site.""Do you have specific wear data? For example, Akron abrasion loss?""Well… we haven’t tested that. The customer feedback says it works better than the previous supplier’s.""How does it compare with other products on the market?""To be honest, we can’t really tell its strengths or weaknesses. Everyone uses similar materials, and the price competition is fierce."
If you have worked in robot sole procurement or R&D in the robotics industry, this conversation will probably sound familiar. This is the real situation in the current field: there is no unified test method, no recognized evaluation criteria, and each enterprise has its own definition of "good" and "durable". The buyer cannot clearly state what they need, the seller cannot explain where their advantages lie, and the entire industry is stuck in an inefficient discourse system.
This is no trivial issue. When an emerging industry has not established a performance evaluation system for its basic components, information asymmetry will drive the bad money to drive out the good. High-quality products cannot demonstrate their high added value through standard tests, low-cost products muddle through with vague descriptions, consumers cannot choose truly reliable products, and the entire industry consumes resources in low-quality competition. Some people call this "involution", but I prefer to call it the absence of standards.
Evaluating a robot sole by focusing only on wear resistance is far from enough. The functions that a sole must undertake include at least: supporting the weight of the robot, buffering the impact of landing, providing sufficient grip, maintaining stable performance during long-term use, adapting to various environmental conditions, and, for intelligent soles with sensors, accurately sensing the contact state with the ground. This article attempts to systematically sort out the evaluation requirements of these five dimensions and put forward a set of operable testing and evaluation framework.
- Five Core Dimensions: How to Test Robot Soles
2.1 Mechanical Properties – The Structural Foundation
Mechanical properties answer the most fundamental question: as a load-bearing structure, can this sole provide sufficient support and cushioning?
The test for static stiffness is relatively mature. We can refer to the test method for compression properties of vulcanized rubber (GB/T 7757) to measure the deformation of the sole under a specific compressive stress. This indicator is directly related to the attitude control of the robot: excessively high stiffness leads to insufficient cushioning, and the impact force is transmitted upward along the leg structure, which may damage the joint motor; excessively low stiffness leads to excessive Z-direction settlement, which affects the accuracy of gait planning. The static stiffness of soles for industrial inspection robots usually needs to be one to two orders of magnitude higher than that of consumer-grade products, but the specific threshold is currently determined entirely by the experience of engineers.
The real challenge lies in dynamic performance. The lessons from the footwear industry deserve attention: many materials perform well in quasi-static compression tests, but when subjected to millisecond-level landing impact, the rate-dependent hardening caused by the viscoelastic effect will greatly reduce the actual cushioning effect. The sports shoe industry recognized this problem as early as 20 years ago. GB/T 30907-2025 specifically specifies the test method for the shock absorption performance of rubber shoes and sports shoes. It simulates the actual landing conditions through drop weight impact or high-speed servo hydraulic system, and measures the peak impact force attenuation rate and energy return rate. The impact conditions of robot soles are similar to those of sports shoes, but there are also essential differences: robots have more regular gait cycles, higher impact frequencies (the typical step frequency of quadruped robots can reach 2-4 Hz), and more predictable load spectra. This means that we can fully learn from the test principles of sports shoes, but we need to adjust the load range, frequency range and road surface conditions to match the actual working conditions of the robot.
Academic research has also confirmed the importance of dynamic performance. A research team designed a humanoid robot foot pad with a lattice structure, and verified through a single-leg drop weight test: when falling from a height of 24 cm, the impact force on the robot with the lattice foot pad was reduced by about 22% compared with that without the foot pad. The SoftFoot adaptive sole developed by the research team of Delft University of Technology not only provides a larger equivalent support surface when walking on uneven ground, but also has an energy absorption effect close to that of a pure flexible sole. These studies show that the impact cushioning performance of the sole is not an icing on the cake, but a systematic factor for the overall dynamics of the robot.
The six-component force test concept from the tire industry is also worthy of reference for robot soles. GB/T 39702-2020 specifies the measurement method of longitudinal force, lateral force, normal force and corresponding moments of automobile tires under steady-state conditions. During the walking process of legged robots, the normal impact force, tangential friction force and the resulting moments borne by the sole also determine the stability and efficiency of movement. Domestic research teams have developed a "test platform for mechanical characteristics of foot-soil interaction of legged robots", which obtains the normal force and tangential force values of the foot end during movement through a three-way force sensor, to guide the foot end design and stability control. The standardization of this direction will bring enormous value to the industry.
2.2 Durability Performance – The Time Dimension
Mechanical properties tell you how good the sole is when it leaves the factory, while durability performance tells you how much performance remains after half a year of use. For industrial scenarios, the latter is often more important than the former.
Let’s start with wear resistance. There are two sets of mature test methods in the footwear industry: one is the whole shoe sole abrasion test method specified in GB/T 3903.2-2017, which simulates the wear of the whole shoe in actual use; the other is the rotating roller abrasion machine method specified in GB/T 9867, which is used to determine the abrasion performance of vulcanized rubber or thermoplastic rubber materials. The tire industry commonly uses DIN abrasion loss (refer to ISO 4649) to evaluate the wear resistance grade of tread rubber. These methods have their own advantages and disadvantages: the rotating roller method is easy to operate and has good repeatability, which is suitable for material-level screening, but cannot reflect the complex friction path in actual working conditions; the whole shoe method is closer to real use, but has a long test cycle and difficult variable control. Robot sole wear testing faces a similar dilemma. A feasible approach is: adopt the DIN abrasion method for rapid screening in the material development stage, and design a special "gait simulation abrasion tester" in the product finalization stage to conduct long-term wear tests on standard road surfaces in accordance with the actual step frequency, step length and plantar pressure distribution of the robot.
Fatigue life is the most neglected shortcoming at present. Most soles fail not because they are worn through, but because the initiation and propagation of micro-cracks inside the material under cyclic loading eventually lead to structural fracture. A complete fatigue life test takes weeks or even months, and almost no enterprise is willing to wait under the pressure of the R&D cycle, resulting in the risk being passed on to users. However, this problem is not unsolvable. There is a mature methodological basis for fatigue testing of rubber products. GB/T 1687 specifies the method for determining the temperature rise and fatigue resistance of vulcanized rubber in flex tests, which can be used to predict the durability of rubber products under dynamic flexure. Some testing institutions have launched the "impact compression life prediction service for quadruped robot foot pads", which analyzes the fatigue characteristics and performance attenuation law of materials by simulating the mechanical environment under actual working conditions. It is a completely feasible technical path to systematize and parameterize such methods and formulate accelerated fatigue test standards for robot soles.
Creep performance is also not to be ignored. The permanent deformation of the sole under long-term compressive load will lead to changes in the standing height of the robot and the contact surface of the sole, thereby affecting the motion accuracy. The footwear industry has already developed mature test schemes for this, including indicators such as compression set and dynamic modulus change. These methods can be applied to robot soles with slight modifications.
2.3 Environmental Adaptability – The Scenario Boundary
A robot that performs perfectly in a constant temperature and humidity laboratory may see its performance drop sharply when placed outdoors at minus 30°C in Northeast China, on acid-base ground in chemical plants, or outdoors exposed to the sun in Hainan. The significance of environmental adaptability testing is to calibrate the use boundary of the product.
The core parameters of temperature adaptability are glass transition temperature (the critical point of low-temperature brittleness) and heat deflection temperature (the critical point of high-temperature softening). For robots operating in cold regions, the former is a rigid requirement – below the glass transition temperature, rubber will lose its elasticity and become hard and brittle, and a single landing impact at low temperature may cause the sole to crack directly. The tire industry has conducted very systematic research on the temperature characteristics of rubber materials, and relevant methods can be directly transplanted to the field of robot soles.
The test for aging resistance is more systematic. The rubber industry generally adopts accelerated aging methods, which simulate years of natural aging effect in days to weeks by increasing temperature, ozone concentration and ultraviolet intensity. For ozone aging, GB/T 7762-2014 specifies the static tensile test method, which observes the crack formation time and degree under the conditions of ozone concentration of 50±5 pphm and temperature of 35-40°C. ISO 1431-1 further specifies the dynamic strain method, which is closer to the actual working conditions by applying periodic tension while exposing to ozone. There are also mature test systems for UV aging and thermal-oxidative aging.
At present, these mature test methods are almost not systematically adopted in the robotics industry. The reasons are complex: first, most sole suppliers are small and medium-sized rubber and plastic processing enterprises, and the investment in aging test equipment is large and the cycle is long, so they lack the willingness to implement; second, robot complete machine enterprises often only carry out verification under normal temperature and humidity, completely ignoring the importance of environmental adaptability. The formulation of standards can precisely solve this problem – when test methods and evaluation indicators are written into industry standards, compliance costs become an "entry ticket" for the industry, forcing all enterprises to raise the quality baseline.
2.4 Interface Performance – The "Dialogue" with the Ground
When discussing sole performance, there is a dimension that is easily classified as "others", but in fact its importance is no less than any of the previous ones: the interface behavior between the sole and the ground, mainly including friction characteristics, grip and contact force transmission.
The tire industry has a very deep understanding of this issue. A standard tire performance evaluation system includes far more than just "wear resistance" and "load bearing", but also a series of indicators interacting with the road surface, such as dry and wet grip, rolling resistance, and cornering characteristics. The force and moment test specified in GB/T 39702-2020 is essentially to depict the complex mechanical interaction between the tire and the road surface. The movement of legged robots relies on reliable tangential force transmission between the sole and the ground: insufficient friction will directly lead to slipping, gait disorder and even overturning; excessive friction will unnecessarily increase the motor load and energy consumption. Bionics research has provided many ideas. A team designed four bionic foot ends by analyzing the attachment mechanism of reindeer hooves on ice and frozen soil, which significantly improved the anti-slip performance under extreme terrain. Another team studied the dry adhesion mechanism of gecko-inspired foot pads and developed a robot gripper that can switch the adhesion/desorption state within 0.5 seconds. Progress has also been made in adhesion on wet surfaces. Some studies have revealed the role of capillary force in tangential resistance through finite element analysis, and proposed an enhancement scheme based on a shape-variable soft pad.
The basic test method of friction is not complicated – a standard friction coefficient tester can give the static friction and dynamic friction coefficients on different road surface materials. But the core problem is: the sole undergoes a periodic stick-slip process during actual walking, and the friction coefficient measured statically cannot reflect this dynamic behavior at all. A more practical test scheme is to install the sole on a test platform with normal loading and tangential dragging, simulate a complete gait cycle, record the tangential force-displacement curve, and extract key parameters such as equivalent friction coefficient, stick-slip amplitude and energy dissipation. Such tests have not yet been standardized, but the industry demand is very clear.
2.5 Sensing Performance – The "Nervous System" of Intelligent Soles
Intelligent soles with sensors are no longer a novelty in the industry. From the simplest thin-film pressure switches to multi-channel pressure sensor arrays, the "sensing layer" of the sole has become a core component of the robot's environmental perception system. The introduction of sensors has also brought new evaluation dimensions: sensing accuracy, response speed, and signal stability.
This field has a mature foundation to draw on: insole plantar pressure measurement systems have more than 20 years of application history in the field of rehabilitation medicine and sports science, with relatively mature test methods. The general calibration standard in the industry is that the deviation between the sensor measurement value and the standard pressure gauge shall not exceed ±5%. There are also domestic association standards that specifically regulate the technical requirements and test methods of insole plantar pressure acquisition devices for rehabilitation. However, academic reviews have also pointed out a common problem: existing sensor insoles generally lack systematic calibration, gait-based verification and human research verification.
The sensing test of robot intelligent soles faces more complex challenges. Rehabilitation insoles mainly measure relatively slow changes in plantar pressure distribution, while the sensing system of robot soles needs to achieve rapid response under high-speed impact conditions – the typical requirement is to complete the complete link from pressure change to electrical signal output within a few milliseconds, while ensuring no baseline drift during long-term use.
The response speed determines whether the robot can perceive the ground state at the first moment of landing and quickly adjust the dynamic gait; baseline drift determines whether the sensor can still maintain measurement accuracy after thousands of impacts. These core requirements that cannot be avoided in engineering are basically blank in the existing standard system.
- Evaluation System: There Is No Best Sole, Only the Most Suitable One for the Scenario
With the complete test data of the five dimensions, how do we draw a final evaluation conclusion?
The answer is that there is no one-size-fits-all standard answer. Because different application scenarios have vastly different weight requirements for sole performance:
For industrial inspection robots that walk long distances on rough concrete floors every day, wear resistance and fatigue life are the decisive indicators;
For home service robots operating on wooden floors and carpets, the priority of floor protection and quietness is much higher than wear resistance;
For military special scenarios, the resistance to low-temperature brittleness at minus 40°C and structural stability at 60°C high temperature are the primary considerations;
For underwater or chemical scenario robots, chemical corrosion resistance and sealing performance will override all other requirements.
Therefore, a reasonable evaluation system should never be a fixed scoring formula, but a decision matrix with adjustable weights. The test methods and data benchmarks are objective and unified, but the weight factor of each dimension is completely determined by the use scenario. Standard-setting institutions need to provide a unified "ruler" and benchmark database, but the evaluation model should allow users to flexibly configure according to actual working conditions.
Flexibility does not mean no bottom line. Mandatory minimum requirements must be set for some core indicators – for example, no permanent plastic deformation is allowed under rated load, and wear resistance must meet the basic safety threshold. These "hard thresholds" form the bottom line of industry quality and prevent products from having problems in the most basic safety and reliability.
- A Hierarchical Standard System Idea
Combined with decades of mature experience in the footwear and tire industries, a reasonable robot sole standard system should include at least three levels, without the need to "reinvent the wheel".
First level: Material-level testing. For the elastomer materials used in the sole, directly borrow the mature national standards for rubber and plastic testing. Including: Shore Hardness (refer to GB/T 531), tensile strength and elongation at break (refer to GB/T 528), Akron abrasion/DIN abrasion (refer to GB/T 9867), flex fatigue (refer to GB/T 1687), ozone aging (refer to GB/T 7762), etc. The standard system of this level is very complete, and enterprises can directly implement it.
Second level: Component-level testing. The special test for the sole as a specific component is also the core incremental work of current standard formulation. The content that needs to be clearly defined includes: the size and shape specification of the test sample, the classification standard of loading conditions (load amplitude, frequency, cycle times), the standardized specification of the test road surface (roughness, hardness, material), and the basic technical requirements of the test equipment. The core difficulty of this level is to balance versatility and representativeness – the test conditions must not only truly reflect the actual working conditions of the robot, but also ensure that the test results between different laboratories are repeatable and comparable.
Third level: Complete machine-level verification. The final performance of the sole after being installed on the complete machine must be closed-loop verified through the complete machine test. The current national standard GB/T 44251-2024 Performance and Test Methods for Legged Robots has provided a complete framework for the performance evaluation at the complete machine level. The durability and environmental adaptability tests of the sole can be used as an extension module of this standard and included in the reliability assessment of the complete machine. At the same time, the ongoing ISO/DIS 18646-5 and the released association standard T/ZSRA 003-2024 provide mature references for the formulation of special standards for soles.
- Who Will Drive It Forward?
Standard formulation is never a purely technical issue. It involves the interest coordination of the upstream and downstream of the industrial chain, the investment cost of test equipment, and the practical consideration of who pays for "compliance". But precisely because it is difficult, someone needs to take the first step.
From the perspective of driving mechanism, the core driving forces come from four directions:
Leading complete machine enterprises: They have the strongest motivation to promote the establishment of standards. A unified standard allows high-quality suppliers to stand out, and greatly reduces procurement costs and quality control risks;
Universities and research institutions: They are the natural main body of test method research and development, and can provide scientific methodological support for the standard system;
Testing and certification institutions: Standard formulation and implementation are their core business scope, with inherent execution capabilities;
Policy makers: Filling the gaps in industry standards is a core public service function from the perspective of healthy industrial development.
From the perspective of implementation path, the model of "first association standard, then upgraded to industry standard, and finally national standard" has been verified effective many times in China's manufacturing industry. Led by an industrial alliance or industry association, jointly with complete machine enterprises, material suppliers, universities, research institutes and testing institutions, first issue a set of association standards acceptable to all parties, verify and iterate in industrial practice, and promote the upgrading of standards when conditions are mature. This is far more operable than pursuing a "one-step" national standard.
From the implementation timeline, the material-level test standards can directly refer to the existing system without any technical obstacles; the component-level test standards can be completed through 1-2 years of methodological research and multiple rounds of verification; the complete machine-level verification standard can be supplemented in a targeted manner on the basis of the existing GB/T 44251-2024. As long as all parties in the industrial chain cooperate to promote it, a preliminary robot sole performance evaluation standard system is expected to take shape within 2-3 years.
Conclusion
Establishing a robot sole performance evaluation system is neither a big deal nor a trivial matter. It is just a basic part of a robot, but it is related to the healthy development of the entire tens of billions level legged robot industry.
The significance of standards is never "restriction", but to provide a unified ruler for the entire industry. With this ruler, buyers and sellers will no longer talk past each other, high-quality products will not be squeezed out of the market by low-quality inferior products, R&D investment has a clear verification goal, and the industry can jump out of the vicious circle of low-quality involution.
The footwear and tire industries have spent decades establishing a complete system of test methods, paving the way for us. A robot sole is not a shoe sole, nor a tire, but its core test requirements just fall at the intersection of experience from these two industries. Standing on the shoulders of predecessors, we are fully capable of completing this standardization path in a shorter time.
This is infrastructure-level work for the industry, with large investment and slow results, but once completed, the entire robotics industry will benefit. Someone has to take the first step.
本文由 RobotSole 机器人足底技术团队原创 关注我们，获取更多机器人足底技术干货
