| The ongoing development of China's deep space exploration capabilities and the successes of its man-made satellites,manned space flights,and lunar probe make Mars exploration an inevitable next step.A large amount of scientific research has been carried out on the manufacture of lunar soil simulants,in addition to terramechanics-based analysis for exploration by planetary rovers.However,the safety and mobility of these rovers have received insufficient research attention.During planetary exploration,wheeled rovers have many advantages.For example,they have a simple structure compared with legged vehicles,and have demonstrated good performance in unstructured environments.Wheeled rovers can also help astronauts to expand their range of motion and i mprove the efficiency of their work.During planetary exploration missions,knowledge of the soil parameters of unknown terrain is advantageous for wheeled rover performance,particularly in terms of the vehicle drawbar pull and wheel drive torque.Knowled ge of the terrain characteristics also ensures the safe landing of planetary landers,aids the development of traversability prediction criteria and traction control algorithms,and improves our understanding of planetary surface composition.To date,however,no soil or rocks have been acquired from Mars,and no advanced scientific instruments such as those used to analyze lunar soil are available for the direct analysis of soil and rocks from Mars.However,the mechanical properties of Martian soil and rocks have been investigated by several spacecraft: the two Viking landers,the Mars Pathfinder lander with the Sojourner rover,two Mars Exploration Rovers,and the Curiosity rovers.All of these spacecraft were equipped with imaging systems that captured information on the natural surfaces and disturbed materials of Mars.Based on the data obtained from the six aforementioned Mars missions,combined with several Mars analogues,a Mars soil simulant for use in Martian terramechanics tests was designed,and its mechanical properties,including density,cohesion,and angle of internal friction,were measured.The least squares support vector machine(LS-SVM),genetic algorithm(GA),and partial least squares discriminant analysis(PLSDA)were used to estimate the terrain-shearing parameters,pressure-bearing coefficient,and mechanical parameters of lunar soil,respectively,which can be used for traversability prediction,risk assessment,and automatic path planning.This paper summarizes the physical and mechanical properties of the Martian surface inferred from landed spacecraft,such as its surface features,grain shapes in sedimentary deposits,soil shear strength,and chemical constitution.Without any special apparatus designed to measure the physical properties of the soil,soil scoops and wheels were used to access the surface or nearest subsurface(less than 30 cm).However,data derived using scoops or wheels to estimate soil properties have been shown to have a high degree of variation in studies of similar Martian soils.Summarized data on the soil physical and mechanical properties derived from the Mars landed spacecraft indicate soil cohesion values of 0-4 kPa,an internal friction angle of 30-40°,estimated bulk density of 1.0-1.6 g/cm3,and grain density of 2.65-3.0 g/cm3.In accordance with the engineering requirements for the testing of rover instrumentation before a mission launch,these reference data were used to guide the design of the Martian soil analogues.These analogues can be classified by the properties of Mars that they best mimic and the specific engineering aims they address.Many soil analogues have been designed by the National Aeronautics and Space Administration(NASA)and European Space Agency(ESA).We were unable to obtain the real grain size distribution curve because of a lack of actual Martian soil.Hence,we used the knowledge of Martian soil gleaned by the spacecraft and soil analogues to guide the design of the JLU Mars simulant,or Martian regolith simulant.That knowledge includes the source of the analogue,grain size distribution,spectral reflectance characteristics of the analogue,and mechanical properties.Most Martian regolith simulants are made from volcanic rock.Mainland China has numerous sources and types of volcanic rock.After comparison of the chemical constitution,spectral reflectance characteristics,bulk density,and grain size of Martian soil and terrestrial soils,volcanic rocks from Jingyu were selected as the JLU Mars simulant.Its mechanical properties were tested,with the results showing soil cohesion values of 0-1.4 k Pa,an internal friction angle of 37-52°,estimated bulk density of 0.96-1.41 g/cm3,and grain density of 2.67 g/cm3.A method for the online prediction of the terrain-shearing parameters of a wheeled unmanned ground vehicle(UGV)traversing unknown terrain is proposed herein.A trained multiple-output LS-SVM is used to predict those parameters and map the engineering data.The proposed method allows the UGV to autonomously traverse the planetary surface and “feel” changes in the cohesion,internal friction angle,and shear deformation modulus of the soil.The method is based on simplified wheel-soil interaction and requires no information on wheel sinkage.Experiments were performed using a single-wheel soil bin to measure the sinkage,drawbar pull,and torque of a griddle net wheel under different slip ratios(0.2,0.3,0.4,0.5,0.6)and wheel loads of 30 N and 50 N.An additional experiment was performed with a continuous slip ratio ranging from 0.2 to 0.6 and a wheel load of 50 N to validate the method.The experimental results show that the multiple-output LS-SVM model is able to accurately predict the terrain-shearing parameters using the slip ratio,torque,and wheel load.Also,the drawbar pull can be calculated using the predicted terrain-shearing parameters,correlation coefficients,and root mean square error of the measured(0.9107)and predicted(1.8856)drawbar pull values.In combination with terramechanics,the GA was used to estimate the pressure-bearing coefficient.The wheel load,torque,and slip ratio were used as the input parameters,and the GA model was run to predict the coefficients of the lumped pressure-sinkage coefficient and slip sinkage exponent.These coefficients were then used to estimate the wheel sinkage based on a modified Bekker model.When the griddle net wheel was driving on the surface of the JLU-2 lunar soil simulant,the estimated pressure-bearing coefficient was [sk,1n,2n ] = [1017.341,1.112,0.649],and the predicted wheel sinkage was relatively close to its measured value,with a correlation coefficient of 0.9660.When an aluminum wheel was driving on the surface of the JLU Mars 2 simulant,the estimated pressure-bearing coefficient was [sk,1n,2n ] = [1929.586,1.038,0.861],and the predicted wheel sinkage was relatively close to its measured value,with a correlation coefficient of 0.9649.Finally,PLSDA was performed to estimate the mechanical parameters of lunar soil in conjunction with 16 signatures proposed on the basis of the information available on the lunar rover,including wheel load,slip ratio,and wheel sinkage information.In this paper,the terrain parameter of the lunar soil is divided into three kinds of bulk density: soft,normal,and hard.A total of 247 items of experimental data were collected from a wheel-terrain interaction testbed,and then randomly assigned to a calibration dataset and prediction dataset at a ratio of 2:1.Hence,the calibration and prediction datasets comprised 166 and 81 items of data,respectively.An optimized discriminant analysis model was developed using mean-center data pretreatment and 10 optimized combinations of signatures.The degrees of accuracy of the calibration and prediction datasets were 90.96% and 90.12%,respectively.These results demonstrate that signatures combined with the PLSDA method are suitable for estimating the mechanical parameters of lunar soil. |
PDF Full Text Request