1. How can sensor-less force sensing overcome the disadvantages of force sensors in intelligent robots?
Sensor-less force sensing approaches have been developed to overcome the disadvantages of force sensors, such as high cost, additional weight, increased complexity, and susceptibility to noise. These approaches aim to provide robots with the awareness of touch without the need for force sensors. Various studies have introduced sensor-less force sensing methods using techniques like disturbance observer (DOB), Kalman filter, and adaptive Kalman filter. These methods utilize acceleration sensors, position sensors, and low-resolution sensors to estimate and control force with reduced noise. However, challenges such as noise, friction, undesired periodicities, and bandwidth limitations still exist. The integration of SSA with DOB has not been thoroughly analyzed for force sensing. Despite the advancements, achieving human-like tactile sensitivity with high-fidelity force sensing remains a challenge. Further research is needed to enhance the performance and reliability of sensor-less force sensing in intelligent robots.
read more
2. What are the advantages of the proposed method over the conventional method in force sensing and control?
The proposed method has several advantages over the conventional method in force sensing and control. Firstly, it does not require the identification of noise characteristics, as the SSA separates components of a signal based on estimating and decomposing the covariance matrix derived from M lagged replicas of the original signal. This makes the implementation of the SSA algorithm straightforward and allows for easy selection of the embedded dimension M. Secondly, the proposed method achieves fine force estimation at a higher DOB pole of 6280 rad/s (1 kHz), widening the force sensing bandwidth to reach the human tactile sensation bandwidth. In contrast, the conventional method requires a known model of periodicity for the Kalman filter, which can lead to significant variance in high order force responses at transient states. The proposed method directly estimates the periodic component in the periodic force, improving periodicity elimination during transient responses. Overall, the proposed method is a simple-design approach that is useful for realizing wideband and high-performance sensor-less force sensing and force control.
read more
3. What are the advantages of using SSA for signal decomposition in comparison to the DPCE Kalman filter?
The advantages of using SSA (Singular Spectrum Analysis) for signal decomposition in comparison to the DPCE Kalman filter include: 1. SSA is an efficient approach of spectral estimation that can handle arbitrary time series signals, including non-linear or non-stationary signals. 2. SSA is a projective and non-parametric method that separates the original signal into components of trends, periodicities, and noise. 3. SSA algorithm does not require the identification of noise characteristics, making it a straightforward implementation process. 4. The dimension M, which determines the performance of SSA noise separation, can be easily and quickly selected through investigating the performance of SSA offline. These advantages make SSA a more effective and efficient method for signal decomposition compared to the DPCE Kalman filter.
read more
4. How does SSA based DOB achieve superior wideband force sensing?
The SSA based DOB achieves superior wideband force sensing by estimating force using motor speed and current reference. Spectral estimation by SSA effectively decomposes noisy components from force estimation, overcoming bandwidth limitations caused by noise when the DOB pole is increased. This results in remarkably widened bandwidth of force sensing. The force estimation by the DOB is designed to estimate load torque, nonlinear friction, and parameter variations as one state variable. The structure of the DOB includes motor torque coefficient, motor inertia, motor torque current, compensation current, motor's rotary angle, speed, and acceleration. The disturbance torque is estimated using motor current and speed, and the compensation current is calculated based on the estimated disturbance torque. Overall, the SSA based DOB enables robust acceleration control and high-performance force sensing.
read more